CRC LEME
Open File Report 59
ABSTRACT
Strategies and methods for the interpretation of geochemical
data. Discussion paper applied to laterite geochemistry
Grunsky, E.C.
This paper summarizes a systematic approach to analyzing the geochemical
data collected over lateritic terrains in the Yilgarn Block of Western
Australia.
Critical steps in the sequence of data analysis are:
- preliminary data analysis,
- exploratory multivariate data analysis, and
- specific multivariate data analysis and modeled multivariate
analysis.
For the evaluation of geochemical data in laterites of the Yilgarn
Block the following sequence of investigation is recommended:
1. Preliminary Data Analysis
- The use of histograms, box & whisker plots, Q-Q plots, scatter
plot matrix, data ranking;
- Preparation of summary statistical tables;
- Maps of elements with each sample ranked into percentile ranges;
- Elimination of gross outliers
- Investigate outliers for each element: analytical error or atypical
abundance;
- Adjust data for censored values;
- Transformation of data based upon samples below the 95th-98th
percentile;
- Scatterplot matrix for transformed data
- Threshold selection after transformation
2. Exploratory Multivariate Data Analysis
- Robust estimates to compute means and covariances to enhance
the detection of outliers;
- Application of dimension reducing techniques such as principal
components analysis;
- The use of methods to delineate structure in the data (cluster
analysis, multi-dimensional scaling, non-linear mapping, and projection
pursuit);
- The use of chi-squared plots applied to transformed data to
isolate outliers based upon all of the elements of interest; maps
of large Mahalanobis distances (>95th percentile) may identify
anomalous areas.
3. Specific Multivariate Data Analysis and Modeled Multivariate
Analysis
- Calculation of empirical indices specifically tailored to areas
in which multi-element associations are understood.
- Multiple regression applied to areas where a linear model of
the multi-element association can be computed with good results
(i.e. high R2 coefficients). Residuals can be examined for the
potential of being associated with mineral deposits.
- The establishment of background and target groups that characterized
the geochemical variation of the regional geochemistry and the
mineral deposits.
- Analysis of variance and canonical variate analysis to test
the statistical uniqueness of the groups.
- The use of all possible subsets to compare reference groups
with each other and determine which group of elements enhance
the group separations.
- The application of allocation-typicality procedures to test
unknown samples from a regional exploration; programme. Each sample
is assigned a probability of belonging to one of the reference
groups. Maps of typicality or posterior probability can be made
to indicate group membership.
This approach of systematically analyzing the laterite geochemistry
forms the basis of an effective exploration programme strategy in
geochemistry.
Last updated: Thursday, January 06, 2000 10:57 AM
|
