Abstract

  Risk is by no means a simple concept. Natural variability and definitional problems with the concept of probability complicate the measurement and use of risk as an analytical tool. Variability requires that risk assessment methods separate natural from total risk when attempting to estimate anthropogenic risk. Failure to do so results in the over estimation of anthropogenic risk and the eventual loss of credibility for risk assessment methodologies. The common frequentist approach to probability is not consistent with anything but a modelling approach to risk assessment. When combined with its ability to account for natural variability, incorporate laboratory-assay data and offer complete statistical and experimental control, modelling is a promising approach to risk assessment. Modelling, however, is not without its drawbacks. Initialization bias can result in the over, or under, estimation of both natural and anthropogenic risk. Furthermore, model estimates are time dependent. The convergence of natural and anthropogenic risk poses problems for modelling-based risk assessment and requires clear statements as to the importance of the time dimension in risk assessment. When combined, the drawbacks to modelling-based risk assessment argue that risk should never be stated as a scalar quantity. Instead, modelling-based risk assessment should provide estimates of the complete range of risk measures (total, natural, and anthropogenic) as well as indications of convergence time. Only then can the modelling-based approach be viewed as the most appropriate means of carrying out scientifically credible risk assessment.

Keywords: risk assessment, natural risk, anthropogenic risk, model, environment