Interpreting exploration geochemical data from Outokumpu, Finland: a MVE-robust factor analysis

Abstract A factor analysis (FA) rendered robust using the minimum volume ellipsoid (MVE) estimator was carried out on geochemical till data consisting of 622 observations on 21 variables (Al, Ba, Ca, Co, Cr, Cu, Fe, K, La, Li, Mg, Mn, Ni, P, Sc, Sr, Th, Ti, V, Y, Zn) from Outokumpu, Finland. The distribution of the majority of variables is slightly to moderately positively skewed. A robust, three-factor model accounting for 82.1% of the total variance was fitted. The fitted model delineates zones related to geology, ore deposits, mineralized horizons, and background variation. The robust FA generally outperforms an otherwise identical classical FA with clearer definition of lithology, background variation, and, in particular, greatly sharpens the anomaly contrasts over mineralization. Both classical and robust factor scores successfully define regional patterns with respect to geology, mineralization and surficial environments. However, both the number and magnitude of scores exceeding 2.0 has increased, from classical FA to robust FA, for all three factors. Such enhancements facilitate the identification of geochemical anomalies, particularly subtle anomalies of the kind commonly associated with till data. This study indicates that the presence of outlying data points can exert a strong masking effect on the identification of geochemical anomalies in classical FA.

[1]  C. Y. Chork An assessment of least median of squares regression in exploration geochemistry , 1991 .

[2]  C. Y. Chork Unmasking multivariate anomalous observations in exploration geochemical data from sheeted-vein tin mineralization near Emmaville, N.S.W., Australia , 1990 .

[3]  David C. Hoaglin,et al.  Applications, basics, and computing of exploratory data analysis , 1983 .

[4]  P. J. Rousseeuw,et al.  Integrating a high-breakdown option into discriminant analysis in exploration geochemistry , 1992 .

[5]  Applications of some statistical techniques to weathered rock geochemical data from the Broken Hill area, New South Wales, Australia , 1991 .

[6]  P. Green,et al.  Analyzing multivariate data , 1978 .

[7]  R. J. Howarth,et al.  Statistical applications in geochemical prospecting: A survey of recent developments , 1984 .

[8]  Ramanathan Gnanadesikan,et al.  Methods for statistical data analysis of multivariate observations , 1977, A Wiley publication in applied statistics.

[9]  C. Y. Chork,et al.  Spatial filtering of exploration geochemical data using EDA and robust statistics , 1989 .

[10]  Hans Kürzl,et al.  Exploratory data analysis: recent advances for the interpretation of geochemical data , 1988 .

[11]  P. Rousseeuw,et al.  Unmasking Multivariate Outliers and Leverage Points , 1990 .

[12]  M. Bernier,et al.  Mineralogical and geochemical analysis of shallow overburden as an aid to gold exploration in southwestern Gaspésie, Quebec, Canada , 1989 .

[13]  John W. Tukey,et al.  Exploratory Data Analysis. , 1979 .

[14]  D. L. Kelley,et al.  Reconnaissance exploration geochemistry in the central Brooks Range, northern Alaska: Implications for exploration of sediment-hosted zinc-lead-silver deposits , 1992 .