Probabilistic risk assessment phase 1. Probability based assessment of environmental and health risks for contaminated land – a review of the literature
Öberg, T.
Report 5532
Swedish Environmental Protection Agency, 2006.

Abstract
Contaminated sites and land result in adverse health effects and environmental risks. This literature review show how the present methodology for quantitative risk assessment can be supported by a probabilistic approach. The probabilistic method involves a characterisation of variability (natural variation) and uncertainty (lack of knowledge). This information can be used to support better decisions and also provide better insight on how to refine the assessment. The probabilistic methodology has become widely used in the United States and is gaining popularity in several European countries.

Probabilistic risk assessments are generally based on simulations of possible outcomes from a large number of possible settings for input variables and model parameters. The calculations can now be performed on an ordinary PC, but some basic skills are required to fully comprehend both the possibilities and the limitations of the methodology.

Contaminated land is together with the nuclear industry the most important environmental application for probabilistic risk assessment, and a substantial number of studies have been published for sites in North America, Europe and Asia. These studies include pollution by lead, arsenic, chromium, uranium, PCB, PAH, hexachlorobenzene, pentachlorophenol, dioxins and chlorinated solvents. The probabilistic risk assessments cover different types of exposure in a diverse set of operations, including metallurgical industry (melting and mining operations), manufacturing industry, gas plants, wood impregnation, infrastructure, and waste landfills.

An increase use of probabilistic risk assessment will require quality assurance procedures and the U. S. EPA has issued a rather detailed guideline document, which also reflects the current scientific consensus. Similar guidelines are needed in Europe.

The need to characterise variability, uncertainty and sensitivity in risk assessment models in Sweden is not different than the need in North America. In addition, the margins of safety must be clearly defined. Probabilistic methods can easily be incorporated into the present Swedish risk assessment models and the report presents an example calculation for benzo[a]pyrene.

Probabilistic risk assessments have often been used to establish site-specific remediation goals and this is projected to an important future application in Sweden. The report suggests that there is a need to establish a framework to simplify the evaluation and interpretation, and recommends that a guidance document be compiled. Training and education is also needed. Courses are already part of the curricula at some universities, but there is also a need for further training and education of working professionals, both by distance education and shorter problem based courses.

A Swedish version of this report is available for download as a PDF file (731 kb)