2008 Joint Meeting of The Geological Society of America, Soil Science Society of America, American Society of Agronomy, Crop Science Society of America, Gulf Coast Association of Geological Societies with the Gulf Coast Section of SEPM

Paper No. 4
Presentation Time: 8:00 AM-6:00 PM

Establishing Background Values in Geochemical Data: Open Source R Language Tools


GARRETT, R.G., Natural Resources Canada - Geological Survey of Canada, 601 Booth Street, Ottawa, ON K1A 0E8, Robert.Garrett@nrcan.gc.ca

Data from geochemical mapping exercises, whether the scale is regional or local, are characterized by two properties - level and relief. Level is a measure of the absolute values in quantification units, and relief is a measure of data heterogeneity, or lack thereof. Areas of complex geology and geochemistry, due to the presence of multiple lithologies and/or the effects of multiple geochemical processes in the primary bedrock and/or secondary surficial environments, are characterized by high relief. The selection of spatial units for which background and relief are to be estimated is an important consideration, to the greatest possible extent they should represent a single entity in the context of the study being undertaken.

Average level is best estimated by the median, a robust measure of central location that is uninfluenced by up to 50% extreme values. However, the average level tells only part of the story. If a question such as, “is a value background?” is to be answered it immediately follows that the range of background has to be known. The estimation of background range is a task that requires the integration of graphical data inspection, the calculation of statistically based estimates, and geochemical knowledge. To support this activity a software package, ‘rgr' has been prepared in the Open Source R language that displays the data in a variety of graphical formats familiar to applied geochemists, and computes a variety of estimates of background range based on non-parametric and traditional and robust parametric estimation procedures. A key display in helping visualize background ranges is the Tukey boxplot. The poster illustrates the various tools available within ‘rgr' and discusses their relative merits.