Assessment of Surface Water Quality in Tank Cascade System: A Case Study in Ulagalla Cascade, Sri Lanka
Analysis and interpretation of the spatial and temporal variation of surface water quality is critical component in
the assessment process. Multivariate analysis such as cluster analysis (CA) and principal component analysis (PCA) were
used to evaluate the spatial/temporal variation and the interpretation of large complex water quality data set to identify the
latent sources of water pollution in Ulagalla cascade, Sri Lanka. The data set included water quality data on 16 parameters for
11 different tanks over a year monitoring period (2016-2017). Based on the spatial variation of water quality along the main
water way, all of the agricultural (pesticides, insecticide and excess fertilizer) and domestic pollutants were accumulated in
the end member of the cascade as water is passing from one tank to another in downstream while it is utilized. Hierarchical
agglomerative clustering was performed to the experimental data standardized through z -scale transformation to minimize
the effect of differences in measurement units. Accordingly, eleven tanks grouped into three clusters based on their
similarities, corresponding to tanks of less polluted (LP), moderately polluted (MP) and polluted (P). PCA performed on water
quality data of eleven tanks indicates that the parameters responsible for water quality variation are primarily related to
mineral dissolution (natural) and factors related agricultural activities as well as livestock. It can be recommended that
multivariate statistical techniques can be successfully used to identify the most important water quality parameters and sites.
Henceforth the results can be effectively use in management of surface water resources in tank cascade systems.
Index Terms - Cluster analysis, mineral dissolution, principal component analysis, Ulagalla cascade, water quality