126
G. Al-Sharrah and H. M. S. Lababidi
validation, which occurs simultaneously with the implementation, and (3) retrospective validation, which occurs after establishment of the methodology and it is usually
applied to real-world data. In this study, the prospective validation of the proposed
approach has been already applied using the random data case presented above (see
Fig. 1).
Three case studies were carried out to demonstrate and validate the proposed
approach. All three case studies are based on actual environmental measurements,
reported in literature, for operational desalination plants in the GCC. The first two
cases studies are considered for the concurrent validation, while the third case study
is for retrospective validation.
4.1 Case Study 1
The first case study is based on data reported by Bu-Olayan and Bivin (2006).
Their study included trace metal levels in seawater from five sites in Kuwait’s
Bay, where many desalination plants exist. The case study covers concentrations
of five metals during harmful and non-harmful algal blooms for both seawater
samples and ctenophore samples in two seasons, summer and winter. The resulted
data matrix consists of five objects (sites I to V) and 40 indicators (concentrations
of five metal in two samples in two seasons in two blooms). All data sets are
provided, i.e. there are no missing data for this case study. Another good point is
that the five metal concentrations were found to be linearly correlated to a high
degree. Hence, data reduction was possible, and it was quite representative to use
the concentration of only one of the metals for the environmental ranking of the
sites. This led to a reduced data matrix of five objects (sites) with eight indicators
(one metal concentration in two samples in two seasons in two blooms). Applying
Copeland ranking for both the original and reduced data matrices gave the same
decisions. The sites were ranked, from the most contaminated to the lowest, as: Site
III (Khadma), Site IV (Towers), Site V(Salmia), Site II (Doha) and Site I (Subiyah).
Hence, excluding dependent indicators was safe given the dependences and large
number of indicators.
4.2 Case Study 2
The next case study is based on data reported by Mohamed et al. (2005) for six
desalination plants in the GCC. Assessment of the plants (objects) is carried out
using 27 indicators, which are listed in Table 3. As shown in the table, there are
several missing data (indicated as ‘NA’), which should be completed if a decision
is to be made about which of the six plants is affecting the environment the most.
Given the relatively high number of indicators, it is recommended to analyze the
data and safely reduce the number of indicators as much as possible. One option is
G. Al-Sharrah and H. M. S. Lababidi
validation, which occurs simultaneously with the implementation, and (3) retrospective validation, which occurs after establishment of the methodology and it is usually
applied to real-world data. In this study, the prospective validation of the proposed
approach has been already applied using the random data case presented above (see
Fig. 1).
Three case studies were carried out to demonstrate and validate the proposed
approach. All three case studies are based on actual environmental measurements,
reported in literature, for operational desalination plants in the GCC. The first two
cases studies are considered for the concurrent validation, while the third case study
is for retrospective validation.
4.1 Case Study 1
The first case study is based on data reported by Bu-Olayan and Bivin (2006).
Their study included trace metal levels in seawater from five sites in Kuwait’s
Bay, where many desalination plants exist. The case study covers concentrations
of five metals during harmful and non-harmful algal blooms for both seawater
samples and ctenophore samples in two seasons, summer and winter. The resulted
data matrix consists of five objects (sites I to V) and 40 indicators (concentrations
of five metal in two samples in two seasons in two blooms). All data sets are
provided, i.e. there are no missing data for this case study. Another good point is
that the five metal concentrations were found to be linearly correlated to a high
degree. Hence, data reduction was possible, and it was quite representative to use
the concentration of only one of the metals for the environmental ranking of the
sites. This led to a reduced data matrix of five objects (sites) with eight indicators
(one metal concentration in two samples in two seasons in two blooms). Applying
Copeland ranking for both the original and reduced data matrices gave the same
decisions. The sites were ranked, from the most contaminated to the lowest, as: Site
III (Khadma), Site IV (Towers), Site V(Salmia), Site II (Doha) and Site I (Subiyah).
Hence, excluding dependent indicators was safe given the dependences and large
number of indicators.
4.2 Case Study 2
The next case study is based on data reported by Mohamed et al. (2005) for six
desalination plants in the GCC. Assessment of the plants (objects) is carried out
using 27 indicators, which are listed in Table 3. As shown in the table, there are
several missing data (indicated as ‘NA’), which should be completed if a decision
is to be made about which of the six plants is affecting the environment the most.
Given the relatively high number of indicators, it is recommended to analyze the
data and safely reduce the number of indicators as much as possible. One option is