Вы находитесь на странице: 1из 31

SEG Global Exploration 2002

Exploration Technology – Discovery Through Innovation

Geochemistry – Introduction
Friday April 12, 2002

L. Graham Closs
Colorado School of Mines

LGC/02
exploration geochemistry or geochemical prospecting

“Geochemical prospecting for minerals … includes any


method of mineral exploration based upon systematic
measurement of one or more property of a natural occurring
material … The purpose of the measurements is the
discovery of a geochemical :anomaly” or area where the
chemical pattern indicated the presence I\of ore in the
vicinity.”
Hawkes, 1957

LGC/02
Components of a GX Program

design and planning


field sampling
sample processing
chemical analysis
data presentation and interpretation

LGC/02
GX Project Design Objectives

optimize target identification


maximize geochemical contrast
minimize false alarm rate
cost effective

Thomson, 1986

LGC/02
a few “simple” assumptions

geological material can be considered to be composite

the challenges is to recognize, isolate and enhance the


“signal” from the individual components (s) of the sample
reflecting both the presence and proximity of
mineralization

the exploration geochemistry process consists of a series


of steps that are both sequential and iterative: the success
of the entire process is based upon the weakest link

LGC/02
basic questions to be addressed by the survey
design and results

for each sample, is there any imprint of a mineralization


process in the geological material sampled and chemically
analyzed – yes or no!

for the subset of “anomalous” samples so identified, are


there any trends or spatial patterns in the data that vector
to a causitive mineralization source

LGC/02
Geochemical Data and Database Management

• Field and Analytical Data must be archived and organized


• Critical information about each survey should be stored with
the data

• Location Coordinates, Type of coordinate conversion


• Project Name, Country
• Field Description Information, Map Sheet
• Type of sample
• Lab Information: Lab name, Date analyzed, Lab Job#.
• Analytical Information: Type of preparation, analysis,
determination method.

LGC/02
Geochemical Reports – Data Management

• magnetic storage media provide convenient, compact


and reliable methods of storing and transmitting data
• floppy disk or CD containing data in pocket of report
• data password-protected to avoid tampering
• scan and store original records, eg. assay certificates
• standard formats recommended: ASCII or RTF
• include both hardcopy listings and electronic data in
report
• store illustrations in high-resolution raster format (tiff
or jpeg formats)

LGC/02
Background, Threshold and Anomaly

• General Definitions:
• background: normal abundance of an element in
unmineralized earth material
• threshold: concentration of an element above which a
sample is considered anomalous
• anomaly: deviation from the norm
Rose et al, 1979

• need to state operational definitions used


• indicate if high values (outliers) excluded, multiple
populations recognized, or data transformations
applied

LGC/02
Threshold Concept

LGC/02
Geochemical Reports – QA / QC

• field geologists quality control program


• components:
contamination, precision and accuracy
• frequency of QC samples
• use of standards and certified reference materials
• participation in comparative round robin programs
• accreditation
• what procedures are used when compliance
standards are not met

LGC/02
Precision vs Problem

LGC/02
Assessing Data – Potential Difficulties
• assumption of the normal distribution of data
• closure problems with weight percent oxide data
• definition of and exclusion of outliers (+/-)
• leveling problems when merging data sets
• data transformations (+/-)
• rules / justification used to address such issues
need to be stated

LGC/02
LGC/02
Data Conditioning
Extreme Outliers
• Mark as missing data
• Delete
• “Cap”, or set to the value at roughly 98% of the
data.
• Log transform (doesn’t always work and may create
negative skew to data).
Standardize (Level)
• Must be used if blending two batches with different
background populations. (Quality control problem, different
soil type, rock type…)
• Caution: Calculations are based on x + 2s and the
assumption of a normal population.
LGC/02
GX Signal Recognition
z Physical Separation:
eg. - 80 mesh soils
z Chemical Speciation:
eg. partial extraction
Fe,Mn oxides
z Geochemical Associations:
eg. pathfinders
As for Au
statistical characterization
z Combinations:
eg. Partial extraction
analysis of - 80 mesh soils
LGC/02
Geochemical Association Concept

LGC/02
Multivariate Statistics I - Correlation
• correlation coefficients – quantify linear relationship
between 2 variables
• transformations or data exclusions clearly identified
• correlation matrices show correlation between all pairs
of variables
• associated scatter plots permit assessment of
assumed linear relationship
• include graphs in text to support interpretations or in
appendices for reference
• initial step in multivariate statistical analysis

LGC/02
Correlation Plots
Scatterplot (MAGUAN.STA 53v*1330c)

5500
y=-42.736+7.7*x+eps
• A few outliers can strongly affect
the correlation coefficient.
4500
R=0.61
3500
CR_PPM

2500
• The correlation coefficient will not
1500
reflect 2 types of correlation within
500
the data.
-500
-20 0 20 40 60 80 100 120 140 160

CO_PPM
• Evaluate the correlation plots not
Scatterplot (MAGUAN.STA 53v*1330c)

650
y=18.021+4.181*x+eps
just correlation coefficients.
550
R=0.76
450 • Condition the data prior to
applying multivariate statistical
CR_PPM

350

methods.
250

150

50

-50
-10 10 30 50 70 90
• Regression Curves defined.
LGC/02
CO_PPM
Data Distributions
Correlation Plot Matrix
Matrix Plot (MAGUAN.STA 53v*1330c)
AL_PCT

AL_PCT CO_PPM CR_PPM CU_PPM FE_PCT MN_PPM NI_PPM


1.00 0.26 0.16 0.02 0.58 0.24 0.30 AL_PCT
CO_PPM 1.00 0.61 0.23 0.83 0.60 0.65 CO_PPM
1.00 0.06 0.45 0.25 0.93 CR_PPM
1.00 0.21 0.20 0.07 CU_PPM
CR_PPM 1.00 0.59 0.55 FE_PCT
1.00 0.36 MN_PPM
1.00 NI_PPM
CU_PPM

FE_PCT

MN_PPM

NI_PPM

LGC/02
Multivariate Statistical Evaluation
Factor Analysis
• Classifies variables
• Defines the structure between variables
• Reduces the number of variables required for
interpretation.
• Each sample is assigned a score for each defined
grouping (factor).

Discriminant Analysis
• Determines which groups of elements discriminate
between 2 naturally occurring groups
(Ore vs. waste)
• Defines the element correspondence for the
groups.
LGC/02
Multivariate Statistics – “Scores”

• ”scores” are weighted combination of individual


elements
• interpreted to represent an optimal measure or
quantification of some process or phenomenon
• assignment of a score to a process is highly
interpretive
• must provide the rules and logic of the interpretation
• score formulae must be clearly defined

LGC/02
LGC/02
LGC/02
Geochemical Data Evaluation

• Correlation between elements


• Interpret geochemical affinities of interest
• Plot Maps
• Proportional Symbol Plots
• Drainage Maps or Drainage Catchment Basins
• Line Plots
• Contour
• Image
• Relief Maps
• Color ternary images

LGC/02
Interpretive Maps
• contour maps
manual: “fact or fiction”
Computer: “more objective”, but …
specify software used
• imaged geochemical data
state contouring parameters
• “score” maps
derived from statistical calculations
derivation of score indicated in legend
criteria used to assign symbols or
contours LGC/02
a few “simple” assumptions

geological material can be considered to be composite

the challenges is to recognize, isolate and enhance the


“signal” from the individual components (s) of the sample
reflecting both the presence and proximity of
mineralization

the exploration geochemistry process consists of a series


of steps that are both sequential and iterative: the success
of the entire process is based upon the weakest link

LGC/02
GX Signal Recognition
z Physical Separation:
eg. - 80 mesh soils
z Chemical Speciation:
eg. partial extraction
Fe,Mn oxides
z Geochemical Associations:
eg. pathfinders
As for Au
statistical characterization
z Combinations:
eg. Partial extraction
analysis of - 80 mesh soils
basic questions to be addressed by the survey
design and results

for each sample, is there any imprint of a mineralization


process in the geological material sampled and chemically
analyzed – yes or no!

for the subset of “anomalous” samples so identified, are


there any trends or spatial patterns in the data that vector
to a causitive mineralization source

LGC/02
GX Signal Recognition
z Physical Separation:
eg. - 80 mesh soils
z Chemical Speciation:
eg. partial extraction
Fe,Mn oxides
z Geochemical Associations:
eg. pathfinders
As for Au
statistical characterization
z Combinations:
eg. Partial extraction
analysis of - 80 mesh soils

Вам также может понравиться