Geography Reference
In-Depth Information
Furthermore, managing and mapping extensive water quality datasets can
be difficult due to the multiple locations, times, and analytes that may be
present. An alternative to numerical modeling is to employ statistical analysis
of groundwater quality data to infer zones of potential contamination.
Principal components analysis (PCA) is a multivariate statistical
technique, which classifies/groups the water quality variables based on their
correlations with each other. The major aim of applying PCA and CA is to
consolidate a large number of observed water quality variables into a smaller
number of factors that can be more readily interpreted. Thus, the multivariate
statistical techniques reduce dimensionality of the data (Dillon and Goldstein,
1984). The PCA helps identifying underlying geologic and hydrogeologic
processes for individual principal components or PCs (or factors) based on the
water quality variables grouped under the PCs. The more PCs extracted, the
greater is the cumulative amount of variation in the original water quality data.
PC loadings show how the PCs characterize strong relationships (positive or
negative) between groundwater quality variable and PC describing the
variable. In order to determine the number of PCs to be retained, Kaiser
Normalization Criterion (Kaiser, 1958) is used.
PCs, which best describe the variance of analyzed groundwater quality
data (eigenvalue > 1) and can be reasonably interpreted (Harman, 1960), are
accepted for further analysis.
The measure of how well the variance of a particular groundwater quality
parameter is described by a particular set of factors is known as ‗communality'
(Jackson, 1991). Number of variables retained in principal components or
communalities is obtained by squaring the elements in PC matrix and
summing the total within each variable. Ideally, if a PCA is successful,
number of PCs will be small, communalities are high (close to 1) and PCs will
be readily interpretable in terms of particular sources or process (Dunteman,
1989). The PCA has previously been used to generate accurate maps of
monitoring wells grouped by their water quality characteristics (Suk and Lee,
1999; Ceron et al., 2000; Güler et al., 2002).
Suk and Lee (1999) performed multivariate statistical analysis in
combination with GIS to correlate contaminant data with groundwater quality
parameters for the purpose of identifying contaminated aquifer zones.
Cluster analysis (CA) is another multivariate statistical analysis technique
that results in data reduction and that can be used to group monitoring sites
according to aquifer water quality behaviour (Suk and Lee, 1999). The CA is
an unsupervised pattern recognition technique that uncovers intrinsic structure
or underlying behaviour of a dataset without making a priori assumption about
