Introduction to Contaminated Site Management - Dealing with Contaminated Sites: From Theory Towards Practical Application - page 36

'supportive input parameters' or the concentrations in contact media. By measur-
ing the organic matter content of soil (a 'supportive input parameter'), for example,
clearly the reliability of a site-specific Risk Assessment can be improved, as com-
pared to an assessment based on an average organic matter content of a specific
region (elimination of variability).
A specific type of uncertainty relates to the lack of clear definitions of polit-
ical boundary conditions or the wrong interpretation of these political bound-
ary conditions by scientists. When, for example, the degree of precaution (e.g.,
whether the average human being or the great majority of human beings must
be protected) has not been clearly defined, or is incorrectly interpreted, the
input parameter identification of exposure parameters could take on an arbitrary
character.
1.5.4.2 Dealing with Uncertainties and Variability
As was mentioned in Section 1.5.4.1 , each input parameter for Human health,
Ecological, and Groundwater-related Risk Assessment is characterised by uncer-
tainty and variability. Nevertheless for many Risk Assessment applications it is
useful to represent the input parameter by one single value. Options that are
mostly used, depending on the type of Risk Assessment, the purpose of the Risk
Assessment, and the possible political boundary conditions, are based on a single
value of the central tendency or some kind of worst-case estimate. Most often, a
single value for the central tendency is the mean (or average) value or the median.
Generally speaking, for normally distributed data the arithmetic mean is appropri-
ate, when for non-normal data the medium value usually is the best representative
of the central tendency. The worst-case estimate is mostly based on a specific per-
centile (usually 80th, 90th, or 95th percentile), or on the highest value found in a
series of data. Although the choice for a specific percentile is also subjective, the
use of a percentile is preferred over the use of an arbitrary high value. In many Risk
Assessments, no specific choice for the level of precaution is made; instead, rather
arbitrary values are selected on the basis of available data in the literature.
Most outputs from site-specific Risk Assessment, such as calculated human
exposure or the number of ecological species affected, must be regarded as indi-
cations of truth values. Nevertheless, Risk Assessment is an extremely useful tool,
as long as it is smartly used.
First, outputs from Risk Assessments can always be safely used for comparison
of risks ( comparative Risk Assessment, aka: relative Risk Assessment ), for example,
for priority setting. Higher exposure, for example, generally means a higher risk;
or to put it even better, a higher Risk Index generally means a higher risk. Second,
Risk Assessments based on worst-case assumptions can be used in a first step of
a Risk Assessment procedure. Generally speaking, this implies that when there is
no unacceptable risk, even under these worst-case conditions, it is relatively safe
to state that unacceptable risks to human health, the ecosystem, the groundwater or
agricultural products are very unlikely. The risk assessor, however, needs to be alert
to the fact that the worst-case conditions indeed apply to the specific site. Imagine,