Impact prediction, evaluation and mitigation - Introduction to Environmental Impact Assessment - page 143

Uncertainty in EIA predictive exercises can be handled in several ways. The
assumptions underpinning predictions should be clearly stated (Voogd 1983). Issues of
probability and confidence in predictions should be addressed, and ranges may be
attached to predictions within which the analyst is n per cent confident that the actual
outcome will lie. For example, scientific research may conclude that the 95 per cent
confidence interval for the noise associated with a new industrial project is 65-70 dBA,
which means that only 5 times out of 100 would the dBA be expected to be outside this
range. Tomlinson (1989) draws attention to the twin issues of probability and confidence
involved in predictions.
These twin factors are generally expressed through the same word. For example, in the
prediction “a major oil spill would have major ecological consequences”, a high degree
of both probability and confidence exists. Situations may arise, however, where a low
probability event based upon a low level of confidence is predicted. This is potentially
more serious than a higher probability event with high confidence, since low levels of
confidence may preclude expenditure on mitigating measures, ignoring issues of
significance. Monitoring measures may be an appropriate response in such situations.
It may also be useful to show impacts under “peak” as well as “average” conditions
for a particular stage of a project; this may be very relevant in the construction stage of
major projects.
Sensitivity analysis may be used to assess the consistency of relationships between
variables. If the relationship between input A and output B is such that whatever the
changes in A there is little change in B, then no further information may be needed.
However, where the effect is much more variable, there may be a need for further
information. Of course, the best check on the accuracy of predictions is to check on the
outcomes of the implementation of a project after the decision. This is too late for the
project under consideration, but could be useful for future projects. Conversely, the
monitoring of outcomes of similar projects may provide useful information for the project
in hand. Holling (1978), who believes that the “core issue of EIA is how to cope with
decision-making under uncertainty”, recommends a policy of adaptive EIA, with periodic
reviews of the EIA through a project's life cycle. Another procedural approach would be
to require an uncertainty report as one step in the process; such a report would bring
together the various sources of uncertainty associated with a project and the means by
which they might be reduced (uncertainties are rarely eliminated).
5.3 Evaluation
5.3.1 Evaluation in the EIA process
Once impacts have been predicted, there is a need to assess their relative significance to
inform decision-makers whether the impacts may be considered acceptable. Criteria for
significance include the magnitude and likelihood of the impact and its spatial and
temporal extent, the likely degree of the affected environment's recovery, the value of the
affected environment, the level of public concern, and political repercussions. As with
prediction, the choice of evaluation method should be related to the task in hand and to
the resources available. Evaluation should feed into most stages of the EIA process, but