Hierarchical Monte Carlo modeling with S-distributions: Concepts and illustrative analysis of mercury contamination in king mackerel

• Published in: Environment international 1995-01, Vol.21 (5), p.627-635
• Main Authors: Voit, Eberhard O., Leonard Balthis, W., Holser, Ronald A.
• Format: Article
• Language: English
• Subjects: Agnatha. Pisces; Animal, plant and microbial ecology; Applied ecology; Biological and medical sciences; Ecotoxicology, biological effects of pollution; Effects of pollution and side effects of pesticides on vertebrates; Fundamental and applied biological sciences. Psychology; Marine; scomberomorus cavalla
• ISSN: 0160-4120; 1873-6750
• DOI: 10.1016/0160-4120(95)00067-U

The quantitative assessment of environmental contaminants is a complex process. It involves nonlinear models and the characterization of variables, factors, and parameters that are distributed and dependent on each other. Assessments based on point estimates are easy to perform, but since they are unreliable, Monte Carlo simulations have become a standard procedure. Simulations pose two challenges: They require the numerical characterization of parameter distributions and they do not account for dependencies between parameters. This paper offers strategies for dealing with both challenges. The first part discusses the characterization of data with the S-distribution. This distribution offers several advantages, which include simplicity of numerical analysis, flexibility in shape, and easy computation of quantiles. The second part outlines how the S-distribution can be used for hierarchical Monte Carlo simulations. In these simulations the selection of parameter values occurs sequentially, and each choice depends on the parameter values selected before. The method is illustrated with preliminary simulation analyses that are concerned with mercury contamination in king mackerel ( Scomberomorus cavalla). It is demonstrated that the results of such hierarchical simulations are generally different from those of traditional Monte Carlo simulations.