Uncertainty estimation to enable better site investigation and risk assessment

1. Uncertainty in measurements of contaminant concentration from site investigation is inevitable, and needs to be quantified 2. It is not sufficient to assume that the samples taken are 'representative', and therefore to ignore the uncertainty generates in the sampling process 3. Similarly, it is not prudent to assume that a chemical analysis is 'correct' if it comes from an accredited laboratory, but better to be advised or to estimate the uncertainty that is generated by the lab, both in the chemical analysis and the physical preparation of the samples that precedes it. 4. The aim should not be to eliminate or even to minimize the uncertainty in an investigation, but rather to know its value and to calculate the optimal value of uncertainty that will achieve fitness-for-purpose in the measurements. This can be the achieved by minimization of the financial loss. 5. The values of measurement uncertainty have several other useful functions, which are: a. To report to the user of the measurements so that they can be propagated through any subsequent calculations (e.g. for risk management) b. The uncertainty is often caused predominantly by the contribution from the small-scale heterogeneity of the spatial distribution of the contaminant. Information on the heterogeneity at a range of spatial scales can be used to improve the estimation of plant uptake of contaminants. This in turn can be used to improve estimates of human exposure. c. Knowing the uncertainty in the original measurements of hazard (contaminant) concentration can be used to better quantify the uncertainty in estimates of exposure and risk to human health. For example if the component of the uncertainty from sampling is found to be dominant, steps can be taken to reduce this level of uncertainty, and thereby improve the reliability of the estimates of exposure and risk