|
|
SummaryDuring the last decades the Danish counties have put a lot of effort into characterisation of contaminated sites and developing methods
for clean up activities. The list of contaminated sites are, however,
increasing faster than the economic resources allocated to the activities
and thus the need for prioritisation of the effort are intensified.
An important challenge in this context is to evaluate the relative importance
of the different parameters used to characterise the contaminated
sites.
Most priority systems for contaminated sites are based on scoring
systems where the influence of the parameters are quantified by developing
e.g. parameter score points and weighing factors and aggregate
these into a risk index. The strength of a scoring based system is
that it is possible to include restrictions into the risk index, a priori,
by setting parameter score values. The weakness is that a linear relation
among the parameters is anticipated even though the parameters
are fundamentally different in origin. Furthermore, the final aggregation
functions used to obtain an aggregated risk index manipulate
the relative parameter influence complicating the interpretation of the
risk index. It is, however, possible to localize unintended influence by
independent methods and calibrate scoring systems in order to optimise
the system with respect to evaluation of the risk potential.
In this project a scoring system, developed by the County of Copenhagen,
"Compound specific priority of point sources" is evaluated in
order to optimise the parameter quantification. The general influence
on the priority index of the intended and unintended priority restrictions
is analysed including a mapping of the influence of the characterisation
parameters. Suggestions on elements in the priority system
that can be optimised are given.
An independent evaluation method is developed by the Danish National
Environmental Research Institute based on partial order theory.
This approach is, unlike scoring systems, not based on the use of
a unified quantification scale for evaluation of the parameter influence.
Partial order theory is used to develop a methodology where
the contaminated sites are compared two by two on each parameter.
By determining technical assumptions the prioritisation results in a
network - opposite to a linear range - but are "projected" by a probability
function to a linear scale.
In general the evaluation shows that the parameter quantification and
structure of the scoring system has both intended and unintended
influence on the risk index. E.g. the parameter weighing factors are
not expressed proportionally in the risk index and the influence of
the individual parameters strongly depends on the structure of the
scoring system.
Two compound specific parameters, i.e. mobility and degradation
and an administrative parameter expressing site classification with
respect to importance as drinking water supply was found to influence the ranking index the most. If the parameter influence was not
weighted a priori the most important ones would be two hydrogeological
parameters and an administrative parameter, i.e. the oxidative
potential of the groundwater, infiltration rate of rainwater and
site classification with respect to importance as drinking water supply.
If the parameter influence is not intended, i.e. may be explained by
environmental knowledge, the scoring system must be calibrated to
an unbiased index. This can be done by replacing the weighing factors
by a calibration factor.
The parameters are grouped into three different risk indexes in the
scoring system: one expresses the risk potential of the geological and
hydrogeological parameters, one the administrative restrictions and
one the compound specific risk potential. For contaminated sites
where characterisation is based solely on historical data a parameter
expressing the probability of detecting any contamination is further
included in the compound risk index. The three indexes are finally
aggregated into a risk index. The compound index has proven to influence
the risk index the most due to both the classification and the
score values of these parameters. The hydrogeological index, however,
that are much more differentiated, is not being expressed fully
in the risk index.
A comparison of a top 10 list with the highest ranked localities determined
with each of the two ranking methods showed a match in 4
out of 10 localities. A further examination of these localities shows a
tendency that the scoring system upgrades heavy metal pollution.
The evaluation has shown that there are both intended as well as
unintended effects of the structure of the scoring system on the
ranking order result. Furthermore unintended effects increase with
the complexity of the system. This means that despite the scoring
system conceptually is easy to understand the results can be difficult
to interpret. Thus it becomes clear that is not quite simple to include
specific priority restrictions into a scoring system. Due to the resulting
uncertainty in the priority results an evaluation of the system is of
interest and contributes to strengthen the reliance in the use of the
results of the scoring system.
|