A Robust Statistical Analysis Approach for Pollutant Loadings in Urban Rivers

Abstract

With rapid urbanization and economic development, anthropogenic activities have brought stressors on urban water resources. This study aims to develop a robust statistical approach for analyzing pollutant loadings in urban rivers to support water management decisions and practices. In order to test the developed approach and demonstrate its feasibility and robustness, a case study was conducted in two urban rivers in eastern Canada. The results indicated that the changes of lead (Pb) concentration in both rivers were not statistically significant among different sites over years; the upward or downward monotonic trend of Pb concentration in each site was also not significant; no significant step trend was found by using Mann-Whitney test; interval analysis verified that the Pb concentration in 2009 and 2010 met the local surface water quality guideline; the self-purification capacities of the two rivers were much limited to reduce the concentrations of Pb in the water; Moreover, the adaptation of the probability plotting method shows the robustness and effectiveness in investigating multiply censored water quality datasets rather than simple substitution procedures commonly used in practice.

Keywords: pollutant loadings, two-way ANOVA analysis, trend analys is, interval analysis, bootstrapping method, multiply censored data analysis