Impact prediction, evaluation and mitigation - Introduction to Environmental Impact Assessment - page 136

( Journal of American Institute of Planners 1965). Mathematical models can also be
divided into deterministic and stochastic models. Deterministic models, like the gravity
model, depend on fixed relationships. In contrast, a stochastic model is probabilistic, and
indicates “the degree of probability of the occurrence of a certain event by specifying the
statistical probability that a certain number of events will take place in a given area
and/or time interval” (Loewenstein 1966).
There are many mathematical models available for particular impacts. Reference to
various EISs, especially from the USA, and to the literature (e.g. Bregman &
Mackenthun 1992, Hansen & Jorgensen 1991, Rau & Wooten 1980, Suter 1993, US
Environmental Protection Agency 1993, Westman 1985) reveals the availability of a rich
array. For instance, Kristensen et al. (1990) list 21 mathematical models for phosphorus
retention in lakes alone. Figure 5.2 provides a simple flow diagram for the prediction of
the local socioeconomic impacts of a power station development. Key determinants in the
model are the details of the labour requirements for the project, the conditions in the local
economy, and the policies of the relevant local authority and developer on topics such as
training, local recruitment and travel allowances. The local recruitment ratio is a crucial
factor in the determination of subsequent impacts.
An example of a deterministic mathematical model, often used in socio-economic
impact predictions, is the multiplier (Lewis 1988), an example of which is shown in
Figure 5.3. The injection of money into an economy—local, regional or national will
increase income in the economy by some multiple of the original injection. Modification
of the basic model allows it to be used to predict income and employment impacts for
various groups over the stages of the life of a project (Glasson et al. 1988). The more
disaggregated (by industry type) input-output member of the multiplier family provides a
particularly sophisticated method for predicting economic impacts, but with major data
requirements.
Statistical models use statistical techniques such as regression or principal components
analysis to describe the relationship between data, to test hypotheses or to extrapolate
data. For instance, they can be used in a pollution-monitoring study to describe the
concentration of a pollutant as a function of the stream-flow rates and the distance
downstream. They can compare conditions at a contaminated site and a control site to
determine the significance of any differences in monitoring data. They can extrapolate a
model to conditions outside the data range used to derive the model—e.g. from toxicity at
high doses of a pollutant to toxicity at low doses—or from data that are available to data
that are unavailable—e.g. from toxicity in rats to toxicity in humans.