A novel approach for indentifying uncertainties within environmental risk assessments

Uncertainties can manifest within the different aspects of environmental risk assessments, affecting the validity of the risk estimate and, in turn, weakening the basis for risk management actions. This research investigated the issues associated with uncertainty characterisation and identification in environmental risk assessments. This led to the creation of a defensible typology of uncertainties, and the creation and validation of a novel uncertainty identification system (UnISERA), based on the elicited views of experts regarding the levels (i.e. magnitudes), natures (i.e. reason for existence) and locations (i.e. where manifest) of uncertainties present within different risk domains. The developed typology, drawn from an analysis of existing assessments, contained seven locations of uncertainty (data, language, system, extrapolation, variability, model and decision), with 20 related sub-types. The output from UnISERA, based on 19 aggregated elicitations across three risk domains (genetically modified higher plants, particulate matter and pesticides), showed that: the risk characterisation phase of assessments contained the highest magnitudes of uncertainty (the level dimension); uncertainties across all four phases of assessments existed primarily through a combination of lack of knowledge and randomness (the nature dimension); and data uncertainty was dominant in the first three phases, and extrapolation uncertainty in the final phase (the location dimension). In comparing the output from UnISERA to similarly produced results in the risk domain of engineered nanomaterials, the nature of uncertainty showed the highest degree of validation (90%), followed by the location (80%) and level (55%) dimensions. The novel approach to uncertainty characterisation and identification presented here will be of use during environmental risk assessments and uncertainty analyses, promoting an understanding of potential uncertainties, and allowing risk analysts to perform assessments with prioritised uncertainties in mind.