Introduction to Contaminated Site Management - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 37

for example, that a worst-case Risk Assessment is performed, for the purpose of
investigating whether people in a specific residential setting might experience unac-
ceptable risks, by using upper limit estimates for the crucial exposure parameters.
If the result from this Risk Assessment under these worst-case conditions shows a
Risk Index that does not exceed a value of 1, an unacceptable risk is very unlikely
for normal conditions at the site. However, when inhabitants at the residential site,
now or in the future, grow a much larger percentage of their vegetables on the site
(large gardens), unacceptable risks cannot be excluded.
Another possible pitfall is that the boundary condition 'based on worst-case
assumptions' is a subjective criterion, which is difficult to motivate and communi-
cate. The level of conservatism is rarely concretised in Risk Assessments or, at best,
at the level of subjective terminology such as 'based on worst-case assumptions'.
Moreover, risk assessors sometimes might feel the urge to protect themselves from
false negatives (the assumption that there is no unacceptable risk, when in reality
there is one) which might lead to an unnecessary over-conservatism.
Scientists and regulators usually are looking for a balance between 'to be sure to
be on the safe side', and realism and pragmatism. For this purpose the term 'realistic
worst case' is often used, although this still is a subjective criterion. The use of a
specific percentile, for example, the 90th percentile of each input parameter repre-
senting worst-case conditions, is a more objective criterion. However, the selection
of this percentile is also a very subjective process. Choices for specific percentiles
(usually 80th, 90th, or 95th percentiles) are often mentioned in Risk Assessments,
but are seldom explained.
For more ambitious applications, the risk assessor needs to be aware of the
sensitivities and uncertainties that are involved in the Risk Assessment tools. An
experienced risk assessor needs to use insight when it comes to the most sensi-
tive input parameters. A sensitivity analysis and uncertainty analyses can help to
systematically identify the most sensitive input parameters, model equations, etc.
The risk assessor also needs to be aware of the limitations of the outputs from
Risk Assessment and check these against the purpose of the Risk Assessment.
When the uncertainties are too great, the performance of additional assessments
will be necessary, or the power of the results of the Risk Assessment will have to be
adapted to more modest conclusions, that is, by communicating the restrictions and
uncertainties.
A relatively simple, though quite time-consuming way of dealing with the lack
of reliability, is to follow a probabilistic instead of a deterministic approach .A
deterministic approach, based on point estimates in input parameters and result-
ing in a single value, does not give any information about the variation in that
value. Moreover, since information about the lack of variation is lacking, stake-
holders might get a misleading idea about the accuracy involved. In a probabilistic
approach, input parameter point estimates are replaced by probability density func-
tions, for at least the most sensitive input parameters. The most popular probability
functions are normal, lognormal, cumulative and uniform distributions. Several soft-
ware packages are available, for example, Crystal Ball, to determine the probability
density functions from a series of data.