Implicit in many of the geostatistical techniques developed, is faith that the data are Gaussian (normal) or can be conveniently transformed to Gaussianity. A mining engineer knows however that contamination is present all too often. This paper will make an exploratory analysis of spatial data, graphing and summarizing in a way that is resistant to that contamination. The ultimate goal is variogram estimation and kriging, and the data analytic techniques used reflect this. The pocket plot is a new way of looking at contributions of small regions of points to variogram estimation. The gridded data are thought of as a higher way table and analysed by median polish. The residual table is shown to contain useful information on the spatial relationships between data. Coal ash measurements in Pennsylvania are analysed from this resistant point of view.
[1]
T. H. Starks,et al.
The effect of drift on the experimental semivariogram
,
1982
.
[2]
Douglas M. Hawkins,et al.
Robust kriging—A proposal
,
1984
.
[3]
G. Matheron.
Principles of geostatistics
,
1963
.
[4]
John W. Tukey,et al.
Exploratory Data Analysis.
,
1979
.
[5]
D. Cox,et al.
An Analysis of Transformations
,
1964
.
[6]
Evaluating sulfur and ash distribution in coal seams by statistical response surface regression analysis
,
1970
.
[7]
R. L. Miller,et al.
Statistical Analysis in the Geological Sciences.
,
1963
.
[8]
N. Cressie,et al.
Robust estimation of the variogram: I
,
1980
.
[9]
W. Sheppard.
On the Application of the Theory of Error to Cases of Normal Distribution and Normal Correlation
,
1899
.