Environmental Engineering Reference
In-Depth Information
It is always recommended to clearly explain the choices undertaken at each step,
and the motivations that have justified them. For example: “
bell-shaped member-
ship functions have been chosen since they better represent the Gaussian behaviour
of the model variables
”.
Once the model has been developed, its application with sample data involves
the following steps:
1.
Fuzzify.
Calculate membership grades of the input data through membership
functions
2.
Infer
. For each rule, calculate grade of antecedent fulfillment (application
of connective
and
), and grade of rule fulfillment (application of connective
if ... then
). Aggregate results of all rules using connective
or
3.
Defuzzify
. Apply defuzzification strategy (when scheduled)
10.3 An Ecological Application: Design of a Quality Index
The development of effective indices of ecological quality represents an important
branch of environmental research, as the capability of measuring human disturbance
on natural ecosystems is necessary for effective management (Marchini 2010). How-
ever, there is little acceptability of ecological indices by environmental managers and
even by scientists. The suitability of old and new indices has been called into question
during conferences and on the pages of scientific journals. Although perceived as
objective procedures, ecological indices involve many steps that are based on subjec-
tive expert judgement: variables selection, data transformations, definition of thresh-
olds, etc. (Scardi et al. 2008). Fortunately, there are advanced computational
techniques such as fuzzy logic, able to handle subjectivity and effectively, by quanti-
fying it and manipulating it with mathematical rigour (Shepard
2005
).
This chapter presents the development of a multi-variable index (Marchini, unpub-
lished data) for transitional waters (estuaries, lagoons) using a fuzzymodel. Themodel
follows the Mamdani type, which is the most commonly used for ecosystemmanage-
ment (Adriaenssens et al.
2004
). Transitional waters are highly variable environments,
therefore, distinguishing between natural and human-induced disturbance is problem-
atic. The European Water Framework Directive (2000/60/EC) requires the inclusion
of biological elements, namely phytoplankton, other aquatic flora, zoobenthos and fish
fauna to measure ecological status of transitional waters. Ecological status has to be
expressed by means of five quality classes:
high, good, moderate, poor
and
bad
. This
requirement is extremely difficult to meet. Generally, transitional waters host low-
diversity communities, with dominance of disturbance-tolerant species. For this
reason, methods based on species diversity or sensitivity might be unable to identify
anthropogenic impacts in these environments. Biological metrics for the definition
of ecological status should be (a) ecologically relevant, i.e. their response to distur-
bance should be unequivocal and acknowledged by a large scientific community, and
(b) easy to measure, i.e. measurements should be low-cost, technically easy to perform
