Main content area

Tracing the potential pollution sources of the coastal water in Hong Kong with statistical models combining APCS-MLR

Liu, Lili, Tang, Zhou, Kong, Ming, Chen, Xin, Zhou, Chunchun, Huang, Kai, Wang, Zhiping
Journal of environmental management 2019 v.245 pp. 143-150
coastal water, factor analysis, nonpoint source pollution, point source pollution, pollutants, principal component analysis, regression analysis, rivers, runoff, seawater, spatial variation, statistical models, surface water, temporal variation, water pollution, water quality, China
In this study, variety of statistical methods were performed to reveal the spatiotemporal distribution characteristics of pollutants and parsing pollution sources of the coastal water in Hong Kong. The temporal-spatial distribution characteristics of the water pollution were various among the three distinct areas, which might be ascribed to the different dominant pollution sources. Cluster and network analysis showed preliminary pollution sources in these areas, and also indicated the temporal characteristics of Deep Bay water pollution, which could divided into two parts before and after 2010. According to the principal component analysis/factor analysis results, three factors in Deep Bay, Tolo Harbour and Victoria Harbour could explained 68.72%, 54.87% and 72.28% of the total variances, respectively. The contribution rate of different pollution source on water quality variables in each area had calculated by absolute principal component score-multiple linear regression model. The contribution rate was roughly ranked as: point source pollution > non-point source pollution > overland runoff > river input. It is the first time to combine multivariate statistical methods, network analysis and regression model to profoundly analyze spatiotemporal variation of seawater quality and parsing the pollution sources. This novel analysis method can provide reference for the water quality evaluation and management of other water bodies.