Clemens Reimann

Multivariate outlier detection in exploration geochemistry

A new method for multivariate outlier detection able to distinguish between extreme values of a normal distribution and values originating from a different distribution (outliers) is presented. To facilitate visualising multivariate outliers spatially on a map, the multivariate outlier plot, is introduced. In this plot different symbols refer to a distance measure from the centre of the distribution, taking into account the shape of the distribution, and different colours are used to signify the magnitude of the values for each variable. The method is illustrated using a real geochemical data set from far-northern Europe. It is demonstrated that important processes such as the input of metals from contamination sources and the contribution of sea-salts via marine aerosols to the soil can be identified and separated.

Low-density geochemical mapping and the robustness of geochemical patterns

Geochemical mapping of entire continents and, ultimately, the world is still a challenge for applied geochemists. At sample densities traditionally used for geochemical exploration (1 site per 1–25u2009km2), geochemical mapping of whole continents is logistically extremely demanding and tremendously expensive. The first low-density geochemical surveys (1 site per 200u2009km2) were carried out during the late 1960s in Africa. Later surveys conducted in various parts of the world had sample densities ranging from 1 site per 300u2009km2 to 1 site per 18u2009000u2009km2. Although these surveys were deemed successful by the investigators in defining variations in background element content on a regional scale, the scientific community was sceptical that low-density geochemical mapping was possible and would provide useful information. The main area of criticism centred around the concern that at such low sample densities the resulting maps would not be robust, i.e. if the same area were resampled and remapped, different geochemical patterns would emerge. Different examples from the USA, Europe, China and Africa demonstrate that low-density geochemical mapping will result in stable and robust geochemical patterns at the continental scale. Such maps are urgently needed for a wide variety of applications.

Multielement regional geochemical reconnaissance as an aid to target selection in Irish Caledonian terrains

Abstract A regional geochemical reconnaissance, based upon the minus 150 micron fraction of active stream sediment, has been conducted by the Geological Survey of Ireland (GSI) over Irish Caledonian terrains. Thirty-eight chemical elements of economic and environmental significance (Ag, As, Au, Ba, Br, Cd, Ce, Co, Cr, Cs, Cu, Eu, Fe, Hf, Ir, La, Li, Lu, Mn, Mo, Na, Ni, Pb, Rb, Sb, Sc, Se, Sm, Sr, Ta, Te, Th, U, V, W, Yb, Zn and Zr) were determined by combined instrumental techniques (AAS and INAA). To date, the multielement database for some 2500 samples from three regions covers about 7,500 km2. Data analysis employed resistant non-parametric techniques, or EDA (Exploratory Data Analysis) for objectively selecting outlier values (anomalies) and class boundaries for the distribution of each element. Such techniques are considered more appropriate to the treatment of geochemical data where assumptions about normality are rarely, if ever, achieved. Class selection was based on the resistant selected order statistic as represented in the boxplot. Special EDA mapping symbols, designed to give equal weighting to all data values, were assigned to the relatively wide boxplot class intervals. None of the original data are redundant in the resulting geochemical maps and, in many instances subtle regional variations can be recognized which are directly attributable to lithological variation. Outlier values are directly correlatable with known bedrock mineralization. The methodology was applied to data from two Caledonian terrains: the Inishowen area of northwestern Ireland and the Leinster massif of southeastern Ireland. A great diversity of metalliferous mineralization (Pb, Zn, Cu, Ba, Au, Ag, W, Li, Ta, U, Cr, Ni, Sb) is found in both terrains. The geochemical maps of Inishowen identify the most prospective gold-bearing lithologies quite effectively. In Leinster, the geochemical maps show the close spatial correlation of Li, Ta, U and W with granitic lithologies, the unequivocal volcanogenic signature of Cu, Sb and Au, an ultra-mafic Cr-Ni association and a new possible volcanogenic Zr-Hf association. While effective in identifying known bedrock mineralization, the maps also suggest possible extensions of mineralized zones and new targets for detailed follow-up surveys. The geochemical patterns defined on the maps are also of assistance in constraining metallogenic models. In addition it can be demonstrated that EDA-based techniques offer a viable and effective method of geochemical mapping.