Crustal Geophysics and Geochemistry Science Center

Modeling Near-Surface Processes in Mineral Systems

Task 4: Develop Prototype Quantitative Models

Task Contact: Dennis Helsel (retired)

Task Summary

Task Highlights

Task Products

Task Summary

The primary objectives of this task were to:

  1. Determine the set of variables that are effective for modeling,
  2. Design prototype models for at least two deposit types, as allowed by availability of data,
  3. Contrast the relative importance of climate, mineralogy, and human activity on the footprint of a deposit,
  4. Forecast model results to locations where field data are incomplete, and
  5. Recommend locations, quantities and or types of data to collect in the future so that future models may be improved and important societal goals met.

Quantitative approaches to environmental modeling were narrowed to a few methods applicable to this project's goals.

Incorporation of spatial characteristics with loglinear or other model types were fit to existing data and conceptual models for massive sulfides, resulting in a first-cut attempt to model trace element movement away from a deposit and into the surrounding soils snd water.

Task Highlights

Methods for simple models of trace-element data were tested using existing available data. The models describe percentiles of trace-element occurrence (how likely is it to observe concentrations at specific levels?), given that some proportion of the data are reported as below a detection limit. Methods for handling data below detection (or reporting) limits has long been an issue in geochemical studies. A textbook providing guidance on this issue was completed and published in early 2005. Expanding and updating methods proposed by USGS in the late 1960s, the book provides procedures that will be required throughout the modeling process of this project.

Related Plots
Photo (click on image for full view, <50k) Description
Lead concentration Boxplot Lead concentrations, including many values reported below one or more detection limits, can be modeled and graphed with boxplots.
Zinc Boxplot Differences in zinc concentrations between two geologic zones, as shown with boxplots. The proportion of nondetected data is higher in the Alluvial Fan zone, resulting in a lower boxplot for that group. Variables that split the data into groups with differing concentrations are candidates for inclusion in a model.
Probability Plot Probability plot of trace element data. The center line is a simple model of occurrence probabilities. This model does not fit the data well.
Survival Lead Plot Survival function plots of two groups of lead data. Probabilities of occurrence (y) are plotted against concentrations (x). Note the concentration scale goes from right to left for these plots. Again, variables that differentiate groups of concentrations are candidates for inclusion in a model.

Return to top of this page | Previous Task | Next Task | Task List

Mineral Resources Program
Eastern Central GMEG Alaska Minerals Information Crustal Geophysics and Geochemistry Spatial Data