Predictive geochemical mapping using environmental correlation

摘要: Abstract The distribution of chemical elements at and near the Earth's surface, so-called critical zone, is complex reflects geochemistry mineralogy original substrate modified by environmental factors that include physical, biological processes over time. Geochemical data typically illustrated in form plan view maps or vertical cross-sections, where composition regolith, soil, bedrock any other material represented. These are primarily point observations frequently interpolated to produce rasters element distributions. Here we propose application covariate regression modelling predict better understand controls on major trace within regolith. Available datasets (raster vector) representing influencing regolith soil intersected with geochemical a spatial statistical correlation model develop system multiple linear correlations. resolution covariates, which much finer (e.g. ∼90 m pixel) than surveys 1 sample per 10-10,000 km 2 ), carries predictions. Therefore derived predictive models concentrations take continuous landscape representations potentially more informative geostatistical interpolations. Environmental applied Sir Samuel 1:250,000 scale map sheet Western Australia individual describing exposed bedrock. As an example two – chromium sodium. We show approach generates high statistically accurate effective ordinary kriging inverse distance weighting interpolation methods. Furthermore, insights can be gained into controlling concentration, mobility from analysis covariates used model. This extended groups (indices), ratios, isotopes range scales variety environments.