Vous êtes sur la page 1sur 29

Jyrki Parkkinen

Geological Modelling and Resource Estimates, a personal view.


Aalto University October 28, 2015

STATISTICS, GEOLOGY RESOURCE


DATABASE METHODS SCREENING FIXING ESTIMATES

SOFT DATA INTUITIVE SOFT MODEL PROBABILITY


STATISTICS MODEL 2
AIM 2

BASIC & PROBABILITY RESOURCE


HARD DATA HARD
SPATIAL MODEL 1 MODEL VALIDATION
MODEL
STATISTICS AIM 1 AIM 3

DIRECTED REPORTING
IMPLICIT
AUTOMATIC MODEL
DESIGN
Parkkinen: Geological Modelling and Resource
10/27/2015 1
Estimates
INTRODUCTION
This lecture is an instant short course, in techniques I have lately used when modelling and estimating
mineral deposits, a kind of cooking receipt + extra spicing with Applied Indicator Method (AIM).
AIM is a data management and arrangement procedure and it supports major interpolation methods.
To understand it you should have elementary, at best also practical, knowledge on some basics like:
- Databases (here ACCESS)
- Spreadsheets (here Excel)
- Basic and spatial statistics, variography (here SAGE2001), stereography (here STEREO32, V. 1.0.3).
- Mining software (here Surpac), block models and interpolation methods (OK, IK, IDW).
Modelling examples are from two sources: Otanmäki Mine Oy and Sotkamo Silver Oy. Note that they
represent intermediate results of ongoing processes, not anything final or acceptable for evaluation.

I recommend students to find more information on the net.


There are plenty of well done resource reports on mineral deposits. Due to the present standards (JORC,
CRIRSCO, NI 43-101, PERC, etc.) good reports are transparent and very informative as to the
techniques used. You can find them on pages of mining companies or you can search for them with
key words like ”mineral resource estimate” and combined with the name of some known engineering
company like ”SRK”, ”Outotec”, ”Micon”, ”Snowden”, ”Wardell Armstrong”, etc.

There are also good courses on the net. I’d recommend:


https://www.youtube.com/watch?v=bm0cfDyl9BY , an introduction to resource modelling.
Some errors, e.g. IDW is considered an isotropic method though it can be anisotropic as well.
https://www.youtube.com/watch?v=SJLDlasDLEU , a series of courses in variography by Ed Isaaks,
writer of several textbooks. These courses are a promo , too, for SAGE2001, a software by Ed.
Parkkinen: Geological Modelling and Resource
10/27/2015 2
Estimates
WHAT IS SOFT DATA, HARD DATA, SOFTSCREENING
MODEL, PROBABILITY MODEL, HARD MODEL?
SOFT DATA / INTUITIVE STATISTICS: SOFT MODEL:
Visual inspection: Familiarity with data 2D-3D Geology:
Interpreted geodata & maps Lithology, Faults,
(rock-types, geophysics, geochemistry, Zones of alteration
stratigraphy, mineralogy, structural data) Deposit delineation
Data consistency/integrity check (Envelope)
Correlations, causations Domains
Process path: what to do, what to use

HARD DATA /BASIC STATISTICS PROBABILITY MODEL HARD MODEL:


”Consult-proof” database: Structural Parameters 3D deposit model, boundaries
Drill holes: azimuth, dip, deviation, length Block model fixed to samples
Drill core logs (lithology), assays, rqd, etc. Global interpolations Final geologic interpretation,
QA/QC for assaying by AIM with OK or IDWn ”Wireframe”.
Basic descriptive assay statistics (lithology, grades)
Cut off grade and Grade capping (Top cut) inside Envelopes/Domains
Structural measurements
In a Probability Model single blocks are assigned values from 0 to 100.
SPATIAL STATISTICS In multi-element or multi-domain interpolations

Variography: parameters for interpolation


Structural data: param. for interpolation
Parkkinen: Geological Modelling and Resource
10/27/2015 3
Estimates
SCREENING: SOFT MODELS: OTANMÄKI AND VUOROKAS
GTK
Otanmäki regional map NYKÄNEN, GTK
Vuorokas map

PAARMA: Otanmäki & Vuorokas


aeromagnetic map,
faults interpreted PÄÄKKÖNEN: Otanmäki
Geological Map Level +125

Parkkinen: Geological Modelling and Resource


10/27/2015 4
Estimates
SCREENING: STRUCTURAL DOMAINS: VUOROKAS (SOFT)
VuoroW VuoroMW VuoroME VuoroE

VUOROKAS W: DRILL HOLES & ROCK TYPES (HARD)


You need soft models
to arrange and control
hard data.

You need intuitive statistics


to keep your feet
on the ground.

Parkkinen: Geological Modelling and Resource


10/27/2015 5
Estimates
SPATIAL STATISTICS:
ARRANGEMENTS / AIM
SAMPLES  COMPOSITES
VUOROKAS DATABASE:
DRILL CORE SAMPLE TABLE FOR
EIGHT ROCK TYPES CATEGORIES
Insert a column for each rock type.
In each column:
Digitize all intersections of a rock type
by giving a digit 100 when present and
a digit 0 when not present.

OTANMÄKI AND VUOROKAS:


TABLE OF 2 M COMPOSITES
WITH PROBABILITY VALUES FOR
EACH ROCK TYPE CATEGORY
Create a table for composite
samples (here 2 m). In mixed
samples you see the precentages
of rock types present.
Vuorokas Rock types
Amphibolite
Anorthosite
Chlor_schist
Gabbro
Granite
Micaschist
Syenite
Ore
Parkkinen: Geological Modelling and Resource
10/27/2015 6
Estimates
SPATIAL STATISTICS: VARIOGRAPHY
Note that different programs may give differing results.
The only way to judge may be visual inspection – a soft way!

SAGE2001:

SURPAC:
Variogram modelling

Parameters for lithology interpolations / VuoroW


Sage: 176/88/-2 major axis 120 m, (see p. 9)
major/med 1, maj/min 2
(Surpac: 160/90/0, major axis 60 m
major/med 1, maj/min 3)
Parkkinen: Geological Modelling and Resource
10/27/2015 7
Estimates
PROBABILITY MODEL 1: VuoroW lithology, horiz. & vert. Sections /AIM, single IDW3
Vuorokas Rock types
Level +120 Amphibolite
Vertical section location is shown by a
Anorthosite red line on Level +120.
Chlor_schist This is a rough litologic model
Gabbro
Granite
created using one IDW3-interpolation;
Micaschist anisotropy-axes by cyan color
Syenite on page 9.
Ore

Level 0
3D Lithology
STRIKE
N40E
Straight borderlines are typical to
this kind of simple interpolation.
A complex interpolation with
several sets of search ellipsoid
parameters gives a softer and
Cross section more credible result as seen on
pages 11-14.
See red lines
Parkkinen: Geological Modelling and Resource
10/27/2015 8
Estimates
SPATIAL STATISTICS: stereogram and ellipsoids

VuoroW ORE 4 SEARCH ELLIPSOIDS

A Surpac
C Surpac
B Surpac
Z Sage correlo
X sage correlo
Y sage correlo
Z sage2 correlo Anisotropy axes of
X sage2 correlo VuoroW lithology
Y sage2 correlo interpolation
A surface 1 See p. 7
C surface 1
B surface 1
B surface 2 Determined from
C surface 2 composite point
A surface 2 surfaces/Surpac
Points of possible
Axes

Stereographic projection for search ellipsoids,


Ellipsoids in space, from above (upper)
Parkkinen: Geological Modelling and Resource
10/27/2015 Towards North (lower). 9
Estimates
SPATIAL STATISTICS:
E
VUOROKAS DOMAINS: SEARCH ELLIPSOIDS

ME
For each Domain, interpolations were done according
to all defined sets of anisotropy parameters:
VuoroW: four sets of parameters (search ellipsoids),
MW Vuoro MW: two sets, VuoroME: two sets, VuoroE:
three sets (see page 13).
Each interpolation was committed in a succession of
varying axial lengths. Each interpolation produced
block values from 0 to 100.
For each Domain, minimum, maximum and average
block values were calculated.
W For results, see pages 11-13.

Parkkinen: Geological Modelling and Resource


10/27/2015 10
Estimates
VUOROKAS Blockmodel, Domains, Attributes, Probabilities, complex IDW3

Level +112 Level +112

VuoroW Probability
MAX ORE %
< 10
10--30
30--50 VuoroW Probability
50--80 AVERAGE ORE %
> 80 >/= 50

Level +112 Domains ME E


MW

VuoroW Probability
W
MIN ORE %
< 10
10--30
30--50
50--80
> 80

Parkkinen: Geological Modelling and Resource


10/27/2015 11
Estimates
PROBABILITY MODEL 1 B /AIM: VUOROKAS W, MW, ME and E Prob. Models
ME object lines of block faces Block Faces of minimum, maximum, and average
MIN 50 %

E
MW ME
W

MAX 50 %

MIN/AVE/MAX 50 %

AVE 50 %

MIN/AVE/MAX 80 %
Parkkinen: Geological Modelling and Resource
10/27/2015 12
Estimates
PROBABILITY MODEL 1 B: VUOROKAS W Ore bodies / AIM,
Several anisotropy structures on vertical sections.
Ore intersections

VuoroW

VuoroW: Average of
Probability Models
On a vertical section
>/= 80 %
>/= 50 %

VuoroW
Average >/= 50 %
Parkkinen: Geological Modelling and Resource
10/27/2015 13
Estimates
Otanmäki Oy: AIM-arrangements in a blockmodel:
Vuorokas block values on a section: ave_vuorow All attributes of one block

Parkkinen: Geological Modelling and Resource


10/27/2015 14
Estimates
IMPLICIT MODEL and HARD MODEL: OTANMÄKI
Metsämalmi, Hard Model

Otanmäki horizontal section +55


Leapfrog: whole Otanmäki
Leapfrog: Metsämalmi
Metsämalmi, Hard Model

Parkkinen: Geological Modelling and Resource


10/27/2015 15
Estimates
IMPLICIT MODEL: SOTKAMO SILVER HARD MODEL: SOTKAMO SILVER
Drill holes and layers A, B, C, D, E, and F
Hard model
Solid A

Parkkinen: Geological Modelling and Resource


10/27/2015 16
Estimates
IMPLICIT vs PROBABILITY vs HARD MODEL: SOTKAMO SILVER SOLIDS

H.M.
H.M. Solid A
Solid A
H.M.
Solid A

Probability Model
vs Hard Model
Y11500 Zn Probability Model
Vs Hard Model
Y11460 Zn
Y11500 Ag g/t
IMPLICIT Ag 30-100
These sections show how both implicit and probability models are
IMPLICIT AG > 100
good for screening and helpful for the construction of
HARD Ag > 50
a hard model. Unfortunately, they need extra parameters to
reach the reliability of hard models. Unproven or even false
anomalies are to be expected.

Parkkinen: Geological Modelling and Resource


10/27/2015 17
Estimates
SPATIAL STATISTICS : SOTKAMO COMPOSITES AND SURFACE FITTING, HARD MODEL A

Sotkamo Ag 200-max g/t Surpac-surface for composites


1 m composites Structural Suite plane

Surpac Structural Suite creates a summary plane for all possible planes
containing composite samples. Planes can be adjusted e.g. according to
grade classes or cut offs. When such planes intersect, the intersection line
Azimuthal direction (bearing)and plunge are given. This is a way to study
deposit (domain) symmetry: principal symmetry plane may equal to the
domain foliation and the intersection line may equal to the domain lineation. Also
Surpac’s Inquire/Bearing and distance between 2 points allows the construction
of possible axes on a surface as demonstrated on the stereo diagram, next page.
Parkkinen: Geological Modelling and Resource
10/27/2015 18
Estimates
SPATIAL STATISTICS : SOTKAMO HARD MODEL, A VARIOGRAPHY FOR Zn/AIM/OK
APPLIED INDICATOR METHOD / OK
Applied Indicator Method Parameters: Solid A
Grade Class Bearing Plunge Dip Anisot.ratio
Grades 10 000-30 000 g/t max/med max/min max range
Lag 40 m 0--1500 150 -50 55 1.7 15 109
1500--4000 145 -45 60 1.5 13 80
4000--10000 150 -50 45 2.5 18 245
Surpac 10000-30000 150 -60 55 3 35 193

Grades 4 000-10 000


Major Axis A
Lag 60 m

Surpac

Grades 0-1 500


Major Axis A
Lag 9 m

Surpac

Parkkinen: Test for Interpolation Methods.


10/27/2015 19
Sotkamo Silver Deposit.
SPATIAL STATISTICS: APPLIED INDICATOR METHOD PROCEDURES, RESOURCE MODEL
PROBABILITY MODEL 1 PROBABILITY MODEL 2 RESOURCE MODEL
(Deposit outlining) (Deposit fixing) (Grades into Domains)

1. SOFT MODELS 1. HARD MODELS 1. RESOURCE MODELS


Enveloping, Domaining Fixed geology Hard model
2. DATA ARRANGEMENT 2. DOMAINING + Grade Domains
Categories (rock types) Classes (grade) 2. PARAMETERS
Classes (grade) Composites Variography
Composites 3. PARAMETERS Inside Domains
3. PARAMETERS Variography of Domains 3. INTERPOLATION
Variography in S.M. Tectonic/Structural data Ordinary kriging
Tectonic/Structural data StructuralSuite (Surpac) IDW2-IDW3
StructuralSuite (Surpac) 4. INTERPOLATION (in Domains)
4. INTERPOLATION (semi-global) Ordinary kriging
Ordinary kriging IDW2-IDW3
IDW2-IDW3

Parkkinen: Geological Modelling and Resource


10/27/2015 20
Estimates
WHAT IS RESOURCE MODEL, VALIDATION, REPORTING

RESOURCE MODEL VALIDATION REPORTING


Hard model Global validation Follow International standards:
Block model Mean model grade vs Jorc, Crirsco, Ni 43-101, or Perc
Grade interpolation composite mean
Composite samples Block histograms vs Detailed description of procedures
inside hard model composite histograms Mineral Resource Statement
domain(s) Alternate estimation methods
Resource classification
(Soft and hard statistics Local validation
combined!) Quality control, QAQC of assays
In sections: blocks vs drill holes
Validation plots: block grades vs
composite grades
Grade-tonnage curves

Parkkinen: Geological Modelling and Resource


10/27/2015 21
Estimates
RESOURCE MODELLING WORKFLOW, three(+) alternate routes:

ALTERNATE WORKFLOW: APPLIED INDICATOR METHOD ADDED AFTER PROBABILITY MODEL 2

Parkkinen: Geological Modelling and Resource


10/27/2015 22
Estimates
PROBABILITY MODEL 2, Sotkamo Silver Hard Model, Solid A,
Zn grade classes and Probabilities, AIM Resource Model

Grade Class: Grade Class: Grade Class:


Zn g/t Zn g/t Zn g/t
< 1500 1500-4000 > 10000

Zn Probability %
In various grade classes (Domains)

Grade Class:
Zn g/t AIM Resource Model:
4000-10000 Zn g/t Final estimate

Parkkinen: Test for Interpolation Methods.


10/27/2015 23
Sotkamo Silver Deposit.
Sotkamo Silver Resource (Block)model: Zn grade classes.
Red arrow points to a block where Zn grade was estimated
by IDW3. Green arrow points to an AIM (OK-)estimate.

IDW3

IK

OK

AIM

Parkkinen: Geological Modelling and Resource


10/27/2015 24
Estimates
RESOURCE MODEL: Always good to have several models created with different methods!

OK 2 IDW2
AIM/
OK

All values g/t


Density 2.8 ORDINARY KRIGING INVERSE DISTANCE INDICATOR APPLIED
3
m Backtransformed logarithmic values Non-transformed WEIGHTED KRIGING INDICATOR
Class Volume Tonnes Straight Clark Lark OK1 OK2 IDW2 IDW3 IK AIM
1. Measured 774 900 2 169 720 4 428 11 208 6 768 6 111 6 206 6 188 6 198 6 263 6 111
2. Indicated 188 150 526 820 3 120 9 370 5 658 4 719 4 512 4 500 4 501 4 497 4 604
3. Inferred 553 250 1 549 100 3 065 13 028 7 867 4 418 4 329 4 361 4 403 4 936 4 250
Grand Total 1 515 600 4 243 680 3 765 11 637 7 027 5 321 5 312 5 312 5 332 5 560 5 246

Parkkinen: Geological Modelling and Resource


10/27/2015 25
Estimates
RESOURCE MODEL CLASSIFICATION

Number of Samples
Distance to
Closest sample

Sotkamo Zn Resource
Classification:
0 Air
Some CLASSIFICATION CRITERIA: 1 Measured
- Level/Quality of Information 2 Indicated
- Robustness of Hard Model 3 Inferred

- Number of Samples per Block


- Distance of Block Centrum
to the Closest Sample
- Average Distance of Samples
Average Distance
to the Block Centrum
- Structural Criteria: Grade
Continuity

Parkkinen: Test for Interpolation Methods.


10/27/2015 26
Sotkamo Silver Deposit.
RESOURCE MODEL VALIDATION Statistics compared
Method Lognormal Ordinary Kriging Ordinary Kriging Inverse Distance Indicator Applied Zn Composites
Straight backtr. Clark Lark OK1 OK2 IDW2 IDW3 Kriging Indicator M.
- Block grade average vs input average Number of samples 10 777 10 777 10 777 10 733 10 733 10 777 10 777 10 733 10 787 2 825
Minimum value 69.88 252.42 152.44 131.03 69.70 94.78 67.32 780.38 336.51 24.00
- Block grade histogram vs input histogram Maximum value 30 000 206 145 124 490 30 000 30 000 30 000 30 000 18 481 26 054 30 000
- Block grade trends vs input trends /XYZ Mean 3 806.47 11 945.69 7 213.95 5 393.42 5 351.14 5 368.05 5 384.41 5 605.89 5 254.48 5 729.28
Median 2 779.89 7 706.68 4 654.03 4 333.39 4 124.96 4 147.55 4 072.91 4 699.06 4 164.96 3 004.73
- Visual validation against drillhole data Standard Deviation 3 687.55 14 315.88 8 645.30 4 230.74 4 404.58 4 578.40 4 727.97 3 658.40 3 722.58 6 958.58
- Compare with results /alternate methods Coefficient of variation 0.97 1.20 1.20 0.78 0.82 0.85 0.88 0.65 0.71 1.21
Skewness 2.74 4.56 4.56 1.86 1.86 1.92 1.93 1.04 1.34 2.03
Kurtosis 13.68 36.41 36.41 8.11 7.77 7.78 7.70 3.63 5.03 6.77

Samples versus Blockmodels

Average Grade along Y-axis


Zn-composites and block averages,
the latter weighted by tonnages,
at 50 m intervals.

Zn g/t
IK

Parkkinen: Geological Modelling and Resource


10/27/2015 27
Estimates
MINERAL RESOURCE REPORTING / PERC & CRISCO

Parkkinen: Geological Modelling and Resource


10/27/2015 28
Estimates
REFERENCES
Bailey, Eddie 2013, The PERC Reporting Standard 2013.
Lindborg, T., Papunen, H., Parkkinen, J., Tuokko, I., 2015. The Taivaljärvi (Silver Mine) Ag-Au-Zn Deposit in Archean
greenstone belt, Eastern Finland. In Maier, Lahtinen, O’Brien (ed.) Mineral Deposits of Finland, Elsevier.
Parkkinen, J. 2009. Geostatistics in Deposit Modeling: Case Kevitsa. Poster, 9th Finnish Geochemical Meeting 2009, GTK,
Espoo, Finland.
Parkkinen, J. 2014. Testing Surpac Structural Suite. Archives of Sotkamo Silver.
Parkkinen, J. 2015. Test for Interpolation Methods: OKlog, OK, IDW2, IDW3, IK, AIM. Archives of Sotkamo Silver.

ACKNOWLEDGEMENT
I would like to thank Messrs Ilkka Tuokko, Managing director of the Sotkamo Silver Oy and Jouko Jylänki,
CEO of the Otanmäki Mine Oy, for kindly permitting me to use real mine and exploration data for the
visualisation of modelling procedures.

Parkkinen: Geological Modelling and Resource


10/27/2015 29
Estimates