You are on page 1of 62

Applied Exploration

Geochemistry

2. How we look at Data

All data in this presentation are the intellectual property of IMEx Consulting (except where
obtained from the public domain) and must not be reproduced, copied, shared or forwarded
in any form without the express permission of IMEx Consulting
Soil samples - Copper
200

Plan View
58
421
240
42
Could be base of saprolite, interface, etc. 35
26
38
30
38
28
42
34
38 16
46 22
78 38 38
56 34 58
42 29 48
40 22 36 38
46 18 34 42
42 30 46 30
39 48 150 42
36 38 60 46
32 34 34 32
24 48 34 32
30 28 98 28
40 42 180 32
48 34 88 32
20 52 290 28
52 38 270 30
42 210 180 35
38 4600 220 30
35 1500 46 60
44 215 42 56
60 1800 38 68
54 2700 36 110
46 530 26 84
48 2600 37 330
58 1400 52 560
45 1100 22 310
80 1400 IMEx Consulting 20 98 2
Soil samples - Copper
200

Plan View
58
421
240
42
Could be base of saprolite, interface, etc. 35
26
38
30
38
28
42
34
38 16
46 22
78 38 38
56 34 58
42 29 48
40 22 36 38
46 18 34 42
42 30 46 30
39 48 150 42
36 38 60 46
32 34 34 32
24 48 34 32
30 28 98 28
40 42 180 32
48 34 88 32
20 52 290 28
52 38 270 30
42 210 180 35
38 4600 220 30
35 1500 46 60
44 215 42 56
60 1800 38 68
54 2700 36 110
46 530 26 84
48 2600 37 330
58 1400 52 560
45 1100 22 310
80 1400 IMEx Consulting 20 98 3
Soil samples - Copper
200
58
421
240
42
35
26
38
30
38
28
42
34
38 16
46 22
78 38 38
56 34 58
42 29 48
40 22 36 38
46 18 34 42
42 30 46 30
39 48 150 42
36 38 60 46
32 34 34 32
24 48 34 32
30 28 98 28
40 42 180 32
48 34 88 32
20 52 290 28
52 38 270 30
42 210 180 35
38 4600 220 30
35 1500 46 60
44 215 42 56
60 1800 38 68
54 2700 36 110
46 530 26 84
48 2600 37 330
58 1400 52 560
45 1100 22 310
80 1400 20 98
IMEx Consulting 4
You missed an orebody!

12 months later….


Another company found an orebody
on OUR ground!
OUR soils data!!
YOU missed it!!!
YOU’RE FIRED!!!!

IMEx Consulting 5
How do we look at data?

What makes a good geochemical target?


How do you define a geochemical target?
What is an anomaly?
Fundamental geochemical data analysis
Introduction to other ways to look at data

IMEx Consulting 6
Do we need to bother?

Jens Klump,
2018. GSWA

IMEx Consulting 7
What is an anomaly?

Is it a big/high number?


“Irregularity, deviation from the common or natural
order” Shorter Oxford Dictionary
An “excursion” or “departure”

IMEx Consulting 8
How do we look at data?

 Univariant analysis
 Percentiles
 Data presentations – Dot maps, images
 Leveling - Z score
 Bivariant analysis
 Scatter Plots
 Multivariant analysis
 PCA, Discriminant Analysis
 Statistical or Geological?
 Are there other ways of looking at data?

IMEx Consulting 9
Univariant Analysis

Excel spreadsheet sort

IMEx Consulting 10
Histogram

How do we define an
anomaly?

95 99
 ppm cut-off

25 50 75
 Percentile

%
 “Obvious” anomalies

5
1
30 100 300 1000 3000
Cu (ppm)

IMEx Consulting 11
Cut Off - Threshold & percentile

Standard statistical analysis


Inflection points - determine sub-domains
Do these have geological meaning?

IMEx Consulting 12
Univariant Analysis

Potential problems
 Cu values for mixed basalt and sandstone terrain
 Calcrete samples with Partial Leach - Base metal values
plateau
 Aeolian dilution
 Primary -> lithotypes - Cu in basalt; Ni in UM
 Weathering environment -> scavenging of mobile elements

IMEx Consulting 13
Data Presentations

Dot plot - univariant


True representation for point data - soils, rocks
Stream sediment samples more difficult - catchment
Images
Beware the griding!
Linear plot
Soil traverse (linked)

IMEx Consulting 14
Dot Plot

Au
>100ppb

IMEx Consulting 15
Image

IMEx Consulting 16
Linear Plot
Spring Gully 6329900N

1200 ASPPMA
1000 AUPPB
CUPPMA
800
ppm/ppb

NIPPMA
600 PBPPMA
SBPPMA
400
ZNPPMA
200
0
752000 752500 753000
Easting
IMEx Consulting 17
Leveling
Premise - apples with apples
Combine different datasets of streams or soils (e.g.
“correct” for mesh sizes)
BUT data comes from adjacent or overlapping areas
with sampling different geological units - doesn’t
replicate

IMEx Consulting 18
7445000mN

Red – 80# samples


7440000mN

7435000mN

7430000mN

Brown – Unk. samples


7425000mN

7420000mN
475000mE

480000mE

485000mE

490000mE

495000mE

500000mE
460000mE

465000mE

470000mE

IMEx Consulting 19
Your Company’s database

“Our database is our greatest asset”


Is there an orebody hidden in the database?
Or how many?
Partially dependant on data quality
Can you get the data out?
Has the data been adequately analysed?

IMEx Consulting 20
Your Company’s database

Cooke, Northern Star, MEGWA Jun 2015 IMEx Consulting 21


Your Company’s data analysis?

How do you look at your data?


How do you combine geology with analysis?
Can you access all of your data?
“Of course we integrate geology with geochemistry”

IMEx Consulting 22
Case History 1 : Univariant

Bookabie Calcrete Sampling


Target: Proterozoic IOCG, SA
Statistical evaluation performed
Average, maximum, minimum, SD, mode and median calculated
“Anomalous values returned for Au, Pb and Mn”
Au : Mean 0.9ppb, max 15ppb, SD 1.2ppb
Pb : Mean 1.8ppm, max 16ppm, SD 3.8ppm
Co : Mean 7ppm, Max 16ppm
IMEx Consulting 23
Case History 1 : Univariant

Bookabie Calcrete Sampling


“Other anomalous values were recorded yet not significantly
high”
“Analysis shows many of the results are above the calculated
background levels suggesting there is significant anomalism in
the area”
• But what was the detection limit?
• Are these values geologically meaningful?

IMEx Consulting 24
IMEx Consulting 25
IMEx Consulting 26
IMEx Consulting 27
IMEx Consulting 28
IMEx Consulting 29
Case History 1 : Univariant

Bookabie Calcrete Sampling


“The main factor in controlling the degree of anomalism is
thought to be the Tertiary Bridgewater Formation that
underlies much of the licence. This unit is believed to impede
water circulation and therefore does not give a true indication of
elements present in the subsurface”
~2,000 at $50 - >$100,000 wasted!

IMEx Consulting 30
Case History 2 : Univariant

Bookabie MMI Sampling


Drilling of magnetic targets intersected high Ni, Cr
• BK67 - 48-51m 3m @ ~0.31%Ni, 0.64% Cr
0-10m Calcrete, -24m CY, -32m SN, -38m CY, -44m SI, -46m
SN, -48m CY, -51m “CY”
Program of MMI soil sampling completed - Univariant only
Is 0.31% Ni anomalous? Why?

IMEx Consulting 31
IMEx Consulting 32
IMEx Consulting 33
Interpretation – Zn in soils

Random Data –
mean 50 & sd = 10

Dave Lawie, 2018. Chile


IMEx Consulting 34
Signal or Noise?

Dave Lawie, 2018. Chile

IMEx Consulting 35
Case History 3 - Univariant

 How do we use stats

 Rock Sampling
 “Anomalous results (determined as
mean plus twice standard deviation)
have been calculated as.......”
 Standard analysis
 How do we treat <?
 What’s missing?

IMEx Consulting 36
Case History 4 - Bivariant
How we use stats
 Correlation
 e.g. K & Rb, Ca & Sr, Al & Ga
 e.g. Mo & Re, Pb & Ag
 Is it useful?

IMEx Consulting 37
Case History 5 - Uni + Bivariant

 How do we use stats

 January 1989
 Soil samples

IMEx Consulting 38
The Analytical Process
How we use stats
 Models
 Machine learning
 Visualization
 Data Interaction

http://www.vismaster.eu/faq/the-visual-analytics-process/ Dave Lawie, 2018. Chile


IMEx Consulting 39
The Trend or the Outlier?
How we use stats
 Exploration
 Want the Outlier!
 Mining/Milling
 Doesn’t want the Outlier – wants the
Trend

Dave Lawie, 2018. Chile


IMEx Consulting 40
Other ways to view data

 Geochemical data is Geologic


More “spatial” than “statistical”
 Use “low end” stats to interact with data
 Lots of “What ifs?”
 Apply Geology, promote variation – “outliers”
 Large datasets - regional trends and obvious anomalies
 Subsets - local variations and orebodies
 Not all elements behave the same way
 Allow data to “speak”
IMEx Consulting 41
Regional Soils - Zn
 What does this image tell us?
 Is there “anomalism”?
 Variations are regolith and
lithology
 How can we tell?

IMEx Consulting 42
Prospect Lag - “Subset”
Cdppm No Data(log)Agppm(log)

Asppm(log)
4 0 6 0 80

3
Auppb(log)

Bippm(log)
0 .3

10
 Link by East
0 .2

1
20

3
0 .1

0 .3

1
 Fe -> As, Bi, Pb, Sb, W
0 .1

0 .3
10 20 30 10 20 30 10 20 30 10 20 30

Feper(log) Feper(log) Feper(log) Feper(log)


40 6 0 80100

200 3 004 00
Mnppm(log)  Co -> part Fe
W ppm(log)Nippm No Data(log)Coppm(log)

Cuppm(log)
1.5

6
1

4
0 .5

100
20

 Also Cu, Mn, Zn


2

10 20 30 10 20 30 10 20 30 10 20 30

Feper(log) Feper(log) Feper(log) Feper(log)


 Mn correlation in PART
Snppm No Data(log)Moppm(log)
3

Pbppm(log)

Sbppm(log)
2 00

20
1 .5
2

1 00

10

 SMALL grid (500m x 250m)


1

0.5

10

Feper(log)
20 30 10

Feper(log)
20 30 10

Feper(log)
20 30 10

Feper(log)
20 30
has strong Fe and Mn
correlations spatially separate
1 00 1 50
Znppm(log)
1 .5

3
1

50
0.5

10 20 30 10 20 30 10 20 30

Feper(log) Feper(log) Feper(log) IMEx Consulting 43


Snppm No Data(log)Moppm(log)
Cdppm No Data(log)Agppm(log)
0.5 1 1 .5 1 2 3 0 .5 1 1.5 0 .1 0 .2 0 .3

10
10
10
10

20
20
20
20

Feper(log)
Feper(log)
Feper(log)
Feper(log)

30
30
30
30
W ppm(log)Nippm No Data(log)Coppm(log) Asppm(log)
1 2 3 0.5 1 1 .5 2 4 6 20 4 0 6 0 80

10
10
10
10

20
20
20
20

Feper(log)
Feper(log)
Feper(log)
Feper(log)

30
30
30
30
Znppm(log) Pbppm(log) Cuppm(log) Auppb(log)
50 1 00 1 50 1 00 2 00 20 40 6 0 80100 0 .1 0 .3 1 3

10
10
10
10

20
20
20
20

Feper(log)
Feper(log)
Feper(log)
Feper(log)

30
30
30
30

Sbppm(log) Mnppm(log) Bippm(log)


10 20 100 200 3 004 00 0 .3 1 3 10

10
10
10

20
20
20

Feper(log)
Feper(log)
Feper(log)

30
30
30

IMEx Consulting
Fe
Prospect Lag - “Subset”

Mn

44
Other ways to view data

Is “large scale” data analysis flawed?


 No – excellent first pass through data – “Obvious
anomalies” detected
BUT will/can miss excursions and less obvious anomalies
These are “near surface” targets that have been
overlooked  The Fallacy of Deep
These are the orebodies in your company’s database!

IMEx Consulting 45
Other ways to view data

Focus : Local scale anomaly definition - “Excursions /


Departures”
Looking at the oxide – near surface - <100m
Orebodies are local variations NOT district scale changes
(lithology dominant)
Large Dataset -> Subsets required
Don’t need consistent element and analytical suite (e.g.
PCA)
IMEx Consulting 46
Why Iron?
 Metallic elements in soils correlate with Fe, Mn (clay) caused by:
 absorption on iron oxides and clays
 variable and mutual dilution by quartz
 Use Fe to recognise background variability of many elements in soils, streams,
oxide rock...
 Soils derived from different rocks may have different levels of Fe, etc due to:
 Primary Fe differences both absolute and in mineralogic form eg sulphides v silicates
 Hydraulic sorting of quartz from finer more labile Fe containing components in sediments
 Gross differences due to rock type ie granite vs mafic vs limestone vs shale
 Secondary effects - pH buffering by CO3 immobilising Fe derived from weathered SX.

IMEx Consulting 47
Iron may be inappropriate…
 Fe commonly metasomatically introduced during hydrothermal mineralisation.
 Fe added during mineralisation, use as a normaliser will suppress of anomalous signatures over mineralisation!
 Metals from mineralisation will be mobile during weathering  controlled by soil acidity.
 High pH generally causing most metals to immobilise either as precipitates or by absorption onto Fe and Mn
oxides. Thus carbonate formations form Terra Rossa soils, indicative of Fe immobility while siliclastic
sequences with little buffering capacity will be comparatively less relatively enriched in iron and other metals.
 Iron not analysed. Another element(s) can be used as a surrogate(s). Provided the
element:
 would covary with iron in background (if iron were analysed)
 is not involved in a significant way in the target mineralising processes.

IMEx Consulting 48
Iron works!

Pb -63m Pb -2.12 +1.18mm & residual Pb (Fe regression)


IMEx Consulting 49
Dunlap et al, 1983
Linking (Glinking)

Ultramafic

200
1km

100
Ni Ppm(log)
200

Mafic
Ni (ppm)
100

4000 4200
East
40000 60000 80000100000
Fe Ppm(log)
Raw values don’t necessarily reflect lithology
IMEx Consulting 50
Dilution & enrichment

Ultramafic
200
100
Ni Ppm(log)

Mafic

1kg - soil 1kg - sand


100ppm Ni 0ppm Ni
10% Fe 0% Fe
40000 60000 80000100000
Fe Ppm(log) 2kg - sand + soil
50ppm Ni
5% Fe
IMEx Consulting 51
Dilution & enrichment
Transported

Upper Sap

Fe stillstand

10m @ 0.12% Cu Upper Sap

Lower Sap

Bedrock

IMEx Consulting 52
Things that (currently) annoy me….

 Max in hole, bottom of hole, Transported

interface, base of Up Sap


transported, top of bedrock Fe stillstand
 None of these contain this
critical information Up Sap
10m @ 0.12% Cu

Low Sap

Bedrock

IMEx Consulting 53
Soil Samples - Copper
200
58
421
240
42
35
26
38
30
38
28
42
34
38 16
46 22
78 38 38
56 34 58
42 29 48
40 22 36 38
46 18 34 42
42 30 46 30
39 48 150 42
36 38 60 46
32 34 34 32
24 48 34 32
30 28 98 28
40 42 180 32
48 34 88 32
20 52 290 28
52 38 270 30
42 210 180 35
38 4600 220 30
35 1500 46 60
44 215 42 56
60 1800 38 68
54 2700 36 110
46 530 26 84
48 2600 37 330
58 1400 52 560
45 1100 IMEx Consulting 22 310 54
80 1400 20 98
Histogram
 High values
 Blue samples – orebody

95 99
 Gets lost in “cloud”
 More data -> less likely to see

25 50 75
 Cu in sandstone -

%
“background” (~10ppm) thus
10x background ~100ppm

5
1 30 100 300 1000 3000
Cu (ppm)
IMEx Consulting 55
Excursion
“Obvious” anomalies Deviation from the
common order
1000 3000

1000 3000
Cu Ppm(log)

Cu Ppm(log)
300

300
100

100
30

30 100 300 1000 30 30 100 300 1000


Ni
FePpm(log)
ppm/100 Ni Ppm(log)
Fe ppm/100
IMEx Consulting 56
Excursion
 No Fe!
1000 3000

Use Zn, Ni, Cu, etc.


 Cu/Fe ratio – No!
Cu Ppm(log)

 Linking is not novel!


300

Spring Gully 6329900N


100

1200 ASPPMA
1000 AUPPB
CUPPMA
800
ppm/ppb
30

NIPPMA
600 PBPPMA

400 SBPPMA
ZNPPMA
200
0
30 100 300 1000 752000 752500 753000
Ni ppm/100
Fe Ppm(log) Easting
IMEx Consulting 57
Ratios can be misleading…

 “In the oxidized soil profile, Fe can be mobile.


Other ‘immobile’ trace elements (e.g. Sc, V, Co,
Cr) can be used as a proxy for Fe content to
determine the mafic character of the underlying
geology and hence a background copper level”.

IMEx Consulting 58
Soil Samples - Copper
200
58
421
240
42
35
26
38
30
38
28
42
34
38 16
46 22
78 38 38
56 34 58
42 29 48
40 22 36 38
46 18 34 42
42 30 46 30
39 48 150 42
36 38 60 46
32 34 34 32
24 48 34 32
30 28 98 28
40 42 180 32
48 34 88 32
20 52 290 28
52 38 270 30
42 210 180 35
38 4600 220 30
35 1500 46 60
44 215 42 56
60 1800 38 68
54 2700 36 110
46 530 26 84
48 2600 37 330
58 1400 52 560
45
80
1100
1400
IMEx Consulting 22
20
310
98
59
Soil Samples - Copper
Zambia
Deposit A - low-grade mineralization, very poor
continuity at different stratigraphic levels in
dolomite, tillite & shale.
Malachite near base of weathered overburden
Very good soil anomaly – max >1000 ppm Cu &
marked background to anomaly contrast.

Deposit B - 8 Mt @ 2.8% Cu - porous


sandstones & conglomerates.
Leaching to >30m, marked depletion of Cu in
weathered overburden.
Poor soil anomaly – max 280ppm Cu & low
background to anomaly contrast

Reedman, J. H., 1979.


“Techniques in Mineral Exploration”.
Applied Science Publishers, Essex.
IMEx Consulting 60
Interactive Data Analysis example

 Regional soils dataset


 Data broken down into manageable areas
 Data subsetted on Mesh size (apples!)
 Data files stripped of blank columns and redundant data (e.g. BDL, same value)
 Subsets are small (<200 samples) to enable local scale variations to be
highlighted
 Plot Array - Bivariant analysis performed simultaneously on ALL elements
 Fe or Zn or Mn (Ni) used to determine scavenging and as a mafic indicator
 Cr used to refine Cu/Ni anomalies
 Seeking clusters of points rather than single point anomalies

IMEx Consulting 61
Conclusions

 Orebodies are local variations


 Different way of looking at data
 Different way of ranking anomalies
 Quick – 3000-5000 samples/day
 Approach not a technique - The
Pirate Code – guide not a rulebook
 The biggest number may not be the
best!

IMEx Consulting 62

You might also like