Environmental Engineering Reference
In-Depth Information
efl uent discharge). Next, proposed project parameters are added (catchment area affected
by mining, or predicted efl uent discharge quality and quantity) and the model is applied
again to predict changes in environmental quality after mine development. The difference
between both predictions is equal to the project-induced environmental change.
Sensitivity analysis is an integral and, in many respects, the most important compo-
nent of impact evaluation. The sensitivity analysis enables the relative importance of
selected parameters to be evaluated. This is especially important considering the many
uncertainties in input data in most model applications. Usually input data are selected as
'most likely' values. The relative response of outputs to changes in inputs is their sensitiv-
ity. Sensitivity analysis aims to determine the effects of individual factors and their varia-
tions on the overall results of an analysis. By changing the value of an input variable, the
response of a system to new external inl uences is tested.
A simple yet powerful method of predicting average environmental impacts and their
likely range is the Monte Carlo analysis, a form of probabilistic modelling. The Monte Carlo
analysis uses randomly generated sets of model input data, representing input probability
distributions. Each set of possible inputs is used to calculate a set of outputs. Calculations are
repeated hundreds or thousands of times. Outputs are treated as a statistical sample, analyzed
to determine the mean, standard deviation, and coni dence limit. Independent of the input
data, distribution outputs will follow Gaussian distribution (The Central Limit Theorem).
Monte Carlo analysis provides the mean and the standard deviation of predicted outputs.
The standard deviation of predicted outcomes is as important as the mean value, since it ena-
bles coni dence limits of impact predictions to be dei ned.
The model application should rel ect the quality of the model input data. Increased
model complexity often closely relates to increased uncertainty in the input data. Errors in
estimating the model input data are carried over to the model application, producing less
accurate model outcomes. In complex environmental systems, impact predictions become
difi cult; at the same time, i nancial and scheduling constraints often emerge.
Mathematical models are best suited to assess physical-chemical changes, rather than to
predict changes in the biological and human environment. Here other methods such as
ecological (species habitat) and visual modelling, or threshold analysis are preferred.
The model application should
refl ect the quality of model input
data.
Ecological Modelling
The cumulative effects of impacts on species populations or habitats can be examined using
ecological models which represent component processes of natural ecosystems. They provide
a simplii ed representation of dynamic, complex systems which often have many interact-
ing components. As with numerical models, ecological modelling can be time consuming
and often command signii cant commitments in establishing complex processes and related
model input data. Again the accuracy of ecological modelling depends largely on the avail-
able input data and the relationships between variables within the model.
Examples of successful ecological modelling exist mainly in the forestry sector.
Fragmentation of forest has a certain discernible effect on biodiversity, populations of
a particular species or stream bed conditions within the area. Variables considered by
such models might include the effect of habitat loss, genetic isolation, and edge effects.
Ecological models are equally applicable to numerous other ecosystems. Because ecologi-
cal models focus on cause and effect linkages, they can generally differentiate additive and
interactive processes. As such they offer one of the best methods for analyzing cumulative
and interactive environmental change, but ecological models are also some of the more
difi cult models to establish.
Ecological modelling can be time
consuming and often command
signifi cant commitments in
establishing complex processes
and related model input data.
 
Search WWH ::




Custom Search