摘要

To assess the quality of a water environment, an in-depth analysis of temporal patterns of contaminant concentrations in water body should be carried out based on unbiased water quality datasets. In this study, we developed a modified log-linear model to account for non-stationary seasonal variations of contaminant concentrations over multiple periods. The model was applied to analyze temporal changes of the Ammonia Nitrogen (AN) concentration at Middle Reaches of Huaihe River (MRHR) and two major tributaries, Shaying River (SR) and Guo River (GR). The modified model outperformed the original models and fitted the data well with Pearson correlation coefficients ranging from 0.67 to 0.86. Temporal patterns of AN concentrations, loads and sources were identified from 1998 to 2015 in connection to the implementation of Five-Year Plans (FYPs, policies for controlling water pollution) in the Huaihe River Basin (HRB). The results show that the AN concentration experienced a significant decrease. Since FYPs focused on controlling AN point sources, the proportion of AN loads derived from point sources decreased from 48-86% to 1-17% in the MRHR and from 66-92% to 2-56% in the SR and GR. However, rebounds of AN concentration occurred in the first year of each FYP period possibly due to discontinuity of the policy implementation over the transition between two consecutive FYPs. High AN concentration anomalies were found in flood seasons, related to pollution discharge beyond limits and/or irrational regulation of sluices. These results have implications for future pollution control policies in the HRB, particularly, the need to reduce the upper limits of contaminant loads for flood seasons, continuity of the policies implementation, reduction of non-point source pollution, rational sluice regulation and integrated pollution prevention programs. The developed model and approach are applicable to other polluted river basins to facilitate water quality assessment and evaluation of pollution control policies.