Structural Analysis of the National Geochemical Survey of Australia Data

Abstract

The National Geochemical Survey of Australia is a low density geochemical survey that collected catchment sediment samples covering most of Australia. Concentrations of 60 elements from both bottom and top soil as well as two grain size fractions were obtained, of which 51 were used in the analysis presented here. As the data are compositional, a centered logratio (clr) transform was applied to enable the use of standard geostatistical techniques. The method of minimum/maximum autocorrelation factors (MAF), a multivariate spatial decorrelation method, was applied to the clr-transformed data by soil horizon and grain-size fraction to assess its suitability for structure identification in a compositional setting. The results show MAF to be effective for the delineation of geochemical regions. A structural analysis of the factors indicates slight differences in continuity across soil layers and grain fraction which lead to differences in the strength with which regional features appear on interpolated maps.