Application of synthetic data to establish the working framework for multivariate statistical analysis of river...

This study applied multivariate statistical analysis (MSA) to synthetic data simulated by a river water quality model to verify whether the MSA can correctly infer the pollution scenario assigned in the river water quality model. The results showed that when assessing the number and possible locations of pollution sources based on the results of cluster analysis (CA), two instead of three pollution point source were identified when considering the hydraulic variations of surface water. When discussing the principal component analysis (PCA) result, the second principal component (PC2) and the Pearson correlation coefficients among the pollutants should also be considered, which can infer that Cu, Pb, Cr, and Ni are contributed by the same pollutant point source, and Cu is also influenced by another pollutant point source. This result also implies that the solid and liquid partition coefficients (K_d) of pollutants can affect the interpretation of the PCA results, so the K_d values should be determined before tracing the pollution sources to facilitate the evaluation of the source characteristics and potential targets. This study established a working framework for surface water pollution traceability to enhance the effectiveness of pollution traceability.