Dr.

Abani R Samal
• PhD (May 2007) Environmental Resources and Policy (Focus:
CPG, RM- SME, Fellow-SEG Energy and Mineral Resources)
Principal, GeoGlobal,LLC Southern Illinois University Carbondale, USA
Salt Lake City, Utah, USA • Master of Science and Diploma of Imperial College (2000)
Mineral Exploration
www.geoglobal.co
Imperial College, London
[email protected] • Master of Technology (1995) Mineral Exploration
Indian School of Mines, India
Adjunct Associate Professor
• Master of Science (1993) Geology
Department of Mining Engineering Khallikote College (Autonomous), Berhampur University,
University of Utah Orissa, India
In industry since 1996
• Specialization: Geostatistics & mineral deposit studies
• Commodity experience: Au-Ag, Cu-Mo (Au, Ag), Fe, Ti, U-
Th, Diamond & gemstones etc..
• Prior employment history::
RioTinto, Pincock Allen and Holt, Ferro Alloys Corporation(India),
World Geoscience Data Services (India) 1
 Best Practices in
Mineral Resource Estimation
& Reporting
Outline
• Introduction
• Geological interpretation
• Exploratory data analyses
Q&A
• Block modeling & application of geostatistics
Q&A
• Validation
• Resource Classification and Reporting
Final Q&A
Various stages in mining project development

Various stages of mineral project development

Preliminary Advanced Project

Mining
exploration Exploration Construction

Mineral Mineral Mineral

Mineralization
Deposit Resource Reserve

Definition Changes with increase of confidence

Risk decreases
Various Studies
These set of activities are designed for reaching a goal at the end of the stage. The
studies include collection and use of data and information collected so far for
achieving the goal of pre-estimated grade and tonnage at certain desired level of
confidence.

Preliminary Advanced Project

Mining
exploration Exploration Construction

Various Studies
Pre-
Feasibility
Conceptual / PEA Feasibility
study
Study

Definition Changes with increase of confidence

Risk decreases
 Quick Quiz: What is wrong with this statement?

“I want a feasibility study of mineral resource

and reserve for a PEA.”

a) The feasibility study requires much higher levels of rigor and

detailed analyses compared to the Preliminary Economic
Analysis (PEA).
b) The PEA is done prior to the pre-feasibility study, which then
leads to feasibility study.
c) A PEA does not require a feasibility study.
d) All of the above.
Introduction: Terminology
• Mineral Reserve: A Mineral Reserve is the economically
mineable part of a Measured and/or Indicated Mineral
Resource.
• Mineral Resource: A Mineral Resource is a concentration
or occurrence of solid material of economic interest in or

Confidence Increases
on the Earth’s crust in such form, grade or quality and
quantity that there are reasonable prospects for eventual
economic extraction. (CRIRSCO, 2013).
• Mineral deposit: mineral occurrence of sufficient size and
grade that it might, under the most favorable of
circumstances i.e. reasonable economic and technological
conditions in foreseeable future, be considered to have
economic potential.
Internationally acceptable definition
Mineral resource estimation

• Estimation of grades and tonnages in a mineral deposit

• A stepwise process followed to estimate resources

The stepwise process:

Exploratory Block model and Resource /
Field campaign and Geological Reserve
data Resource Reserve
Data Collection Model estimation
analyses classification reporting

9
Resource estimation in various stages of
project development
• As we move from Conceptual Studies to PFS  the confidence
on estimated resources should improve.
• The importance of getting it right also increases.
• The strategy should lead to developing the projects with a goal
to increase confidence on the estimated resources.
• Requires understanding of the Resource Estimation process.
Resource estimation process
•Topography
Data collection and
•Drill hole data: location of holes, traces, geochemical analyses etc..
QA/QC •Geological structures etc..
• Lithological domains
Geology model • Structural domains (faults, fractures, folds etc.)
• Geochemical domains (grade, alteration domains etc.)
• Statistical analyses
Exploratory data • Visual inspection
analyses (EDA) • Geostatistical model (variogram)
• Grade interpolation
Grade estimation • Optimization of estimation parameters should be considered for
increasing confidence on the estimated grade

Resource • Resource classification based on the confidence of the competent

person and input of all stake holders from all disciplines while
classification applying the criteria for 'eventual economic extraction'
• Visual (cross sections), statistical and spatial (swath plots) validation
Model Validation techniques
• Quality of estimated grades etc..

• Report resources based where inferred resources are separate from

Resource Reporting measured and indicated category based on international codes
Exploration data collection
• Topography
• Geology mapping
• Faults / structural maps
• Drill hole data
• Density data

Recovery of core in a coal exploration project

 Geology and Grade model
• Based on Geological control of mineralization
• Lithological domains
• Structural domains (faults, fractures, folds etc..)
• Geochemical domains (grade, alteration domains etc..)
• 1% Cu shell or, high grade Fe domain etc..

Geochemical domains Lithological / Structural domains

 Best practices in geological modelling
1. Differentiate the geological controls of mineralization
(structure, lithology, porosity etc.) vs others
• Post mineralization structures that relocate the ore-bodies vs minor-
structures (swarm of mineralized fractures)
• Lithological units that control mineralization (such as porous sand-stones
for copper) vs. the units that does not play any role in mineralization
2. Avoid unnecessary
fine-tuning and too many
data points ?
3. Avoid extrapolating too 200 m

much
14
An example of grade continuity
1.2% Cu grade shell shell
Improvement of Geological model

based on surface exploration data (1983)

with added underground information (1986)

based on extensive underground development

(1989)
Neves–Corvo copper–tin mine in Portugal (Fig. 2.6; Richards and Sides, 1991)
Source: Applied Mineral Inventory Estimation, Alastair J. Sinclair and Garston H. Blackwell 2004
Exploratory Data Analyses (EDA)
for mineral resource estimation
• Data integrity – Quality and reliability of drill hole data is very
important
• Reliability of assay data
• Geological logs: Trained logger, cross-checking
• Data maintenance: ownership, chain of custody etc..
For this webinar from this point onwards, we assume this part is taken care of.
EDA: Understanding and modeling the statistical structure of the data
that leads to selecting parameters for grade estimation
• Statistical analyses
• Visual inspection
• Geostatistical model (variogram)
Geology model EDA Exploration parameters
Drill hole data analyses

• Assay data
• Assay length
• Bias
• Compositing
• Appropriate composite length
• Geology matters?
• Composite data analyses
• High grade limit
• Variography (Geostatistical analysis)
 Drill hole assay quality
Geology logs may be
biased towards grades (?)
Longer assays with high grade
or
Shorter lengths are of higher grades
Drill hole assay compositing: types
• Straight: Straight compositing is an ASCII dump rather than a
true compositing technique.
• Run Length: This technique attempts to produce composites of
equal length
• Bench; Bench compositing is similar to the Run Length
technique, except that the composite intervals (or composite
lengths) are determined by chopping the drill hole by bench
rather than by downhole lengths.
• Geology: Geology compositing produces composite intervals as
defined by the geology. Consecutive intervals of the same
geologic code are averaged to form a single composite.
(from Vulcan software)
Which composite length to use: Industry practices

1. Guided by bench height (if known)

• Dilution already added
• Half or full bench height
2. Equal to block height
• Assuming all vertical holes
• Use minimum 1 sample during interpolation?
3. Similar to assay length
• Minimize dilution due to compositing
• Represents the in-situ grade
 Contact Profile Analysis or boundary analyses
Purpose: to investigate the
relationship between a chosen
grade variable when moving from
one geological or estimation unit
to another.
Process: samples from one
geological unit are paired with
samples from another geological
unit based on a separation
distance from the contact. Rock 3 Rock 2

The separation distance is

geometric rather than spatial.
Dealing with very high grade data: Outliers?
Outliers: .……that lies outside the overall pattern of a distribution
• In raw drill hole data or, in composited data
Average 0.53 • Outliers (?) are part of the dataset
STDEV 0.81
CV (1) 1.52 • Identify them
Mean - 2 s -1.08 • Treat them as appropriate
Mean + 2 s 2.14
Mean - 3 s -1.88 • Need special treatment – during estimation
Mean + 3 s 2.95 • May need capping or, define a high yield
restriction for interpolation
• Capping: Capping technique replaces the
high grade values with capping grade
• High yield restriction: This technique
reduces influence of high grade values on
resource estimation
Quick Quiz: Select the high grade value carefully
The statistics of a set of drill hole composite data presented below at various
high grade cut limits. Given this is the only information available to you,
which is the appropriate high grade limit you will consider for use?

Possible Answers:
a) 3% Cu
b) 4% Cu
c) 3 to 4% Cu any value
d) 5% Cu
e) 4 to 5% Cu any value
Spatial Autocorrelation and Variography
• Spatial Autocorrelation is a correlation of a variable with itself
through space.
• If there is any systematic pattern in the spatial distribution of a
variable, it is said to be spatially auto-correlated.
• Random patterns exhibit no spatial autocorrelation.

X(i+h)

X(i)
 Measure of anisotropy: Variogram map

• A 2D / 3D representation
of variogram model
• A valuable tool for
anisotropy analysis
• Should be considered as
an essential part of
variogram modeling
Admissible variogram models
g ( h )  c0  c1  
3 h 1 h
g (h)  c0  bh   ( )3  for 0  h  a

 
2 a 2 a 
g ( h)  c0  c1 for h  a g ( h)  c0  c1 1  eh / 
g(h)

h
Linear model

g ( h )  c0  c1 1  e( h /  ) 
g ( h)  c0  bh
2

 

Hole effect model

Refer: Authorized Variogram models.pdf

Important points

• The nugget effect and the slope of the structure are

important.
• The slope can be assessed from the first three or four variogram
values
• the nugget effect can be estimated by extrapolating back to the
origin.
• Sill is set at the value where the variogram stabilizes
Why not to use least squares (LS)
to fit variogram models
1. The model must be a positive definite function
• otherwise the variance could turn out to be negative
2. Least Squares assumes that sample points are independent
• Not correct in case of variogram – the locations are known
3. The behavior of the variogram very close to the origin (h 0)
• The LS techniques do not understand that
Questions?
Block model

• A block: a 3D space

© Abani R Samal, Pincock Allen and Holt, USA

Block model

Blocks represent >= 1% Cu in this example

© Abani R Samal, Pincock Allen and Holt, USA

Estimation (interpolation) Techniques
Non geostatistical Geostatistical
Polygonal Kriging (OK, SK)
Nearest-neighbor IK
Inverse distance MIK
power
Conditional simulations

Deterministic Probabilistic
Polygonal IK
Nearest-neighbor MIK
Inverse distance power Conditional simulations
Kriging (OK, SK)
Polygonal technique
• For simple layered type deposits
• Each hole has a range of influence = half of distance between
this and the next hole

© Abani R Samal, Pincock Allen and Holt, USA

Inverse Distance Power (IDP)
• Weight of sample is inversely proportional to the
distance of the sample from the location of the
unknown value
• Variants: Power of 2 (IDS), 3 etc.
• Simple to apply
• Bull’s eye effect ?

© Abani R Samal, Pincock Allen and Holt, USA

Ordinary Kriging
• Most popular geostatistical technique
• Condition:
• Data meets geostatostical assumptions: second order
stationarity etc..
• Existence of a spatial structural model (variogram)
Best : Least error
Linear :  (wi * xi) /  (wi)
Unbiased :  (wi) = 1
Estimator
Simple to apply: because does not require the paramters such as a “mean”

© Abani R Samal, Pincock Allen and Holt, USA

The geostatistical estimation processes:
Best practices
• Develop a spatial continuity / structure model (such as
variogram)
• Analyze anisotropy using variogram maps
• Use this model in the estimation process (OK or, SK etc.) or,
simulations
• Post process the models – in case of MIK, simulations etc.. non-linear
techniques
• Risk analyses:
• Risk: Uncertainty for which the probability is known or can be inferred
• Confidence limits, errors of estimation, sensitivity due to block size
etc..
An example

Source: Armstrong, 1998, Basic Linear Geostatistics

Where should we use Kriging?
• The kriging technique, being a geostatistical technique, should
be applied only if the underlying assumptions of second-order
stationarity are met, i.e., at a minimum the mean and variance
of the sample data remain invariant in space. If the mean and
variance of the samples do not meet these stationarity
assumptions, it is advisable to apply an interpolation technique
other than kriging.
• Understanding of geology is must for application of geostatistics
• “ . . . . geology should form the foundation of any
geostatistical analysis”
• Krige, DG, 1999, Essential basic concepts in mining geostatistics and their links with geology and
classical statistics” S. Afr. J. Geol, 102(2),147 - 151
Important points about kriging: Advantages
• Range of techniques:
• Simple Kriging: for continuous ore bodies
• Ordinary Kriging: Most popular for large, continuous to relatively
discontinuous ore-bodies
• Indicator kriging: For categorical variables such as lithology types
Indicator kriging:
• Co-Kriging: for using more than one (two at a time) related variables
• (A case study: Integration of different-quality data in short-term
mining planning. COSTA_Cristina Paixão Araújo.pdf)
• Other techniques: MIK, log-normal kringing, disjunctive kriging etc..
• Option to model errors of estimation
• Estimation of confidence intervals
Important points about kriging: Drawbacks
• Linear technique: so can not provide probabilistic estimation of
continuous variables
• Can result in negative weights
• Kriging variance is dependent on robustness of variogram
• SK can bias the results towards the mean used
Estimation quality depends of following parameters
• Domain
• Geological: orientation, extension etc..
• Search parameters
• Orientation
• Size and shape
• Relation to variogram
• Sample Criteria
• Minimum number of samples
• Maximum number of samples
• Maximum number of samples per hole
• Variogram parameters
• Type of model
• Nugget: sill ratio
• 1st range vs. 2nd range
Search ellipse vs. Geology unit

Spherical search

B
Elliptical search
Quick Quiz:
Which search ellipsoid is geologically valid?
A, B, or both?
A

Spherical search

B
Elliptical search
Comparison of grade tonnage curves generated
from different search schemes
Cumulative
60,000
Spherical Search (Stage 1)
Elliptical search (Stage 2)
Elliptical search (Stage 3)
50,000

applied at 15% Zn
High Yield restriction
40,000

Tonnage (Y units)
30,000

20,000

10,000

0
2x 3x 4x 5x 6x 7x 8x 9x 10x 11x 12x 13x 14x 15x 16x 17x 18x 19x 20x >20
Zn %
45
To protect the rights of the owner of the data, the real grade and tonnage numbers are not shown.
Questions?
Reliability of grade estimation
Validation of resource model

1. Visual validation
• Drill holes (assay or, composites) vs. estimated
2. Statistical validation
• Statistical tables
• Histograms of input vs. estimates (vs. real values)
3. Grade Tonnage curves
4. Swath plots – Spatial validation
Visual validation: Cross sections

A deep copper deposit: Cu grade estimates are compared with

the drill hole assay data along a vertical cross-section
Statistical validation
Simple statistical comparisons may provide
a quick but significant validations of models
• Min & Max
• CV CU_NN CU_Opt_IDS CU_COMPOSITES

• Averages Number of samples /

Blocks
258652 258652 11490
Average 0.75 0.73 1.18
Minimum 0.00 0.00 0.00
Maximum 11.91 13.06 16.26
Variance 0.80 0.50 1.61

Median 0.51 0.57 0.84

95% C.I.- 0.74 0.73 1.15

95% C.I.+ 0.75 0.73 1.20
Statistical plots (Histogram)
Block model estimates

Composites

50
Statistical plots (CFP)
Block model estimates

Composites
Swath plots

• Provides more local comparisons

• The name comes from remote
sensing
• The means of the input data and
estimates are compared along a
swath
• Number of samples used and number
of blocks estimated should also be
considered
Swath plots:
0.35

0.3

0.25

0.2
TCu%

0.15

0.1 Average of TCU % in Model 1

Average of TCu values in Blast hole model

0.05

West to East
0
39650 39850 40050 40250 40450 40650 40850 41050
Easting
Grade Tonnage Curves

Question: Is the OK estimate better than IDS estimates?

Grade Tonnage estimation:
Which one is reliable?

[ GeoGlobal,LLC ]
Quick Quiz
Why it is essential to show all charts and tables
used in the validation processes?

a) It is a good idea because it uses a lot of pages.

b) The tables are not readable, Charts will help to read the table
c) Both A and B.
d) None of the above.
Resource classification
International Reporting Standards
CRIRSCO Compliant Standards
CRIRSCO, formed in 1994,under the auspices of the Council of Mining and
Metallurgical Institutes (CMMI)
Members:
1. JORC (Australasia)
2. CBRR (Brazil)
3. CIM (Canada)
4. Comision Minera (Chile)
5. Kazakhstan – KAZRC – Kazakhstan Reporting Code
6. PERC (Europe)
7. MPIGM (Mongolia)
8. NAEN (Russia)
9. SAMCODES (South Africa)
10. SME (United States of America)
CRIRSCO - Principles

Governing principle: transparency, materiality and competence

• Materiality: All relevant information for a reasoned and
balanced judgement
• Transparency: Sufficient information clear and unambiguous in
order that the reader is not misled
• Competency: Based on the work of suitably qualified and
experienced persons who are subject to an enforceable
professional code of ethics
Source: Development_of_CRIRSCO_IGC_2016.pdf
 The core definitions should be considered in conjunction with Figure 1 in the
CRIRSCO Template
CRIRSCO Template.

Figure 1. General relationship between Exploration Results,

Mineral Resources and Mineral Reserves
Definitions of Mineral Resources
• A Mineral Resource is a concentration or occurrence of solid
material of economic interest in or on the Earth’s crust in such
form, grade or quality and quantity that there are reasonable
prospects for eventual economic extraction.(CRIRSCO)
• . . . . . . . in such form, grade or quality and quantity that there
are reasonable prospects for eventual economic extraction.
(SME 2013)
• . . . . . that there are reasonable prospects for eventual
economic extraction. (JORC 2012)
In my interpretation . . .
• There are two different categories of criteria in “reasonable
prospects for eventual economic extraction”
• 1: Reasonable: These are group of technical and economic
criteria that impacts economic extraction
• 2: Eventual: This refers to time related variables such as life of
mine

• Suggested reading: Additional Guidance - Reasonable prospects

for economic extraction – December 15, 2009 (CIM)
 Reasonableness of the continuity of orebody
Is there an
assumption made
regarding continuity
of mineralization /
grade?

Is one version of the 0.25 g/t solids

0.5 g/t solids
model more
continuous ?
Why?
Cowan et al (2003) Practical Implicit Geological modeling, 5th International Mining Geology Conference
Reasonable Process engineering parameters:
The mineral processing
parameters for two different
geological domains should 1
depend on mineralogical
composition of the material 2
under consideration.
Let’s simplify this . . . .
Apply criteria for reasonable
prospect for economic extraction

Is all this mineable?

Optimum Resource
pit outline

Geological inventory
or, downgrade to
inferred?

65
Quick Quiz:
Can the area marked as A and/or B be considered as
resource? An example based on a real case-study

Factors that may limit the

mining activity:
• Mineral inventory in the
A
Himalayas may not be B
feasible for extraction, hence
to become mineral Possible Answers
resources. a) Only A
• Mineralization under b) Not B
inaccessible public land can c) Both A and B
not be mineral resource d) None of the above
Continued…

Cut-off grade of estimated value

Using geological control
in resource estimation
may lead to reduces
ore-waste

True value
misclassifications. O

Hence reducing risk and Cut-off grade of true value

loss in value W

Estimated value

© GeoGlobal,LLC.
THANKS