Dependent Indicators for Environmental Evaluations of Desalination Plants
121
2 Dependent Indicators for Seawater and Brine
Correlation between indicators can be simply identified through expert knowledge
or mathematically using statistical analysis of collected data. Dependent indicators
are not necessarily redundant, and dependencies through correlations do not necessarily mean causations (Guttman 1977), but may represent different concepts.
Hence, exclusion of dependent indicators may affect the overall decision, and it
is possible only when set against the analysis of influence on decision.
In the desalination field, environmental indicators are defined to characterize
water or air releases. For airborne releases, general dependencies include emissions
of greenhouse gases resulting from the required power generation (Younos 2005).
Indicators that are usually considered for assessing the impacts to air pollution are
basically related to the amount and type of the burned fuel. For the waterside, the
indicators are defined to describe the effect of the rejected brine on the marine
environment. Table 1 lists a number of examples on the main dependencies between
environmental indicators. It provides also examples of primary variables that may
be used in correlating the values of dependent indicators.
3 Ranking with Correlated Data
As mentioned earlier, models that are used in correlating environmental data may
be linear, multiple-linear, non-linear, or other more complex forms (Piegorsch and
Bailer 2005). The proposed approach is to study the dependencies in environmental
indicators and utilize these dependencies in refining the available data. The aim
is to improve the environmental decision making when ranking is applied. The
dependencies can be effectively used in completing missing data as well as
excluding redundant indicators. The key challenge is identifying the dependencies
and determining the correlation models.
In this work, the Copeland score (Copeland 1951) ranking method is used. It is
more than a half-century-old voting procedure, which is simply based on pairwise
comparisons of candidates. It is one of many vote-aggregation systems that socialchoice theorists have invented in their attempts to determine the most appropriate
systems for a variety of voting situations. The Copeland rule selects the object with
the largest Copeland score, which is the number of times an object beats other
objects minus the number of times that object loses to other alternatives when
the objects are considered in pairwise comparisons. Using the concept of partially
ordered sets and social choice theory, the Copeland score ranking methodology was
applied outside of its usual political environment (voting) by Al-Sharrah (2011)
to rank objects in science and engineering applications. This method assumes
neither linearity nor any mathematical relationship among indicators and is therefore
defined as a non-parametric method. The method is presented next.
121
2 Dependent Indicators for Seawater and Brine
Correlation between indicators can be simply identified through expert knowledge
or mathematically using statistical analysis of collected data. Dependent indicators
are not necessarily redundant, and dependencies through correlations do not necessarily mean causations (Guttman 1977), but may represent different concepts.
Hence, exclusion of dependent indicators may affect the overall decision, and it
is possible only when set against the analysis of influence on decision.
In the desalination field, environmental indicators are defined to characterize
water or air releases. For airborne releases, general dependencies include emissions
of greenhouse gases resulting from the required power generation (Younos 2005).
Indicators that are usually considered for assessing the impacts to air pollution are
basically related to the amount and type of the burned fuel. For the waterside, the
indicators are defined to describe the effect of the rejected brine on the marine
environment. Table 1 lists a number of examples on the main dependencies between
environmental indicators. It provides also examples of primary variables that may
be used in correlating the values of dependent indicators.
3 Ranking with Correlated Data
As mentioned earlier, models that are used in correlating environmental data may
be linear, multiple-linear, non-linear, or other more complex forms (Piegorsch and
Bailer 2005). The proposed approach is to study the dependencies in environmental
indicators and utilize these dependencies in refining the available data. The aim
is to improve the environmental decision making when ranking is applied. The
dependencies can be effectively used in completing missing data as well as
excluding redundant indicators. The key challenge is identifying the dependencies
and determining the correlation models.
In this work, the Copeland score (Copeland 1951) ranking method is used. It is
more than a half-century-old voting procedure, which is simply based on pairwise
comparisons of candidates. It is one of many vote-aggregation systems that socialchoice theorists have invented in their attempts to determine the most appropriate
systems for a variety of voting situations. The Copeland rule selects the object with
the largest Copeland score, which is the number of times an object beats other
objects minus the number of times that object loses to other alternatives when
the objects are considered in pairwise comparisons. Using the concept of partially
ordered sets and social choice theory, the Copeland score ranking methodology was
applied outside of its usual political environment (voting) by Al-Sharrah (2011)
to rank objects in science and engineering applications. This method assumes
neither linearity nor any mathematical relationship among indicators and is therefore
defined as a non-parametric method. The method is presented next.