Analysis of spatio-temporal variations of river water quality and construction of a novel cost-effective assessment model: a case study in Hong Kong

Abstract

Assessment of river water quality has been attracting a great deal of attention because of its important implications for the living environment of human beings and aquatic organisms. River water quality is commonly assessed using dozens of different water quality parameters. However, different parameters may contain redundant information, which could lead to the waste of monitoring efforts. Thus, this study constructed a novel cost-effective assessment model of river water quality using the 1-year monitoring data collected from 23 sampling stations in the water control zone of Tolo Harbour and Channel in Hong Kong. First, the spatio-temporal variations of water quality parameters and the overall status of river water quality were analyzed based on all 19 parameters using Kruskal–Wallis test, hierarchical cluster analysis, and the water quality index (WQI). The results indicated that most water quality parameters and overall water quality status varied significantly over space, but did not exhibit obvious seasonal differences; and 99.27% of water samples were identified to be in good or excellent status of overall WQI. Then, using principal component analysis (PCA)/factor analysis (FA) and Pearson’s correlation analysis, eight parameters, including 5-day biochemical oxygen demand (BOD₅), chemical oxygen demand (COD), ammonia–nitrogen (NH₃-N), nitrate-nitrogen (NO₃-N), chlorophyll-a (Chl-a), fluoride (F⁻), total suspended solids (TSS), and arsenic (As), were verified to be responsible for the greatest contributions to water quality, the assessment of overall water quality status. These eight crucial parameters were further employed to establish six cost-effective water quality assessment models. Using the overall WQI as the benchmark, the results of linear regression analysis demonstrated that the cost-effective model constructed based on BOD₅, COD, NH₃-N, NO₃-N, F⁻, TSS, and As were the optimal water quality assessment model, which can achieve the most reliable results with reduced parameters.