Long-term (2003−2021) evolution trend of water quality in the Three Gorges Reservoir: An evaluation based on an enhanced water quality index

• Published in: The Science of the total environment Vol. 915; p. 169819
• Main Authors: Sang, Chong, Tan, Lu, Cai, Qinghua, Ye, Lin
• Format: Journal Article
• Language: English
• Published: Netherlands Elsevier B.V 10.03.2024
• Subjects: Chlorophyll-a; Machine learning; Three Gorges Reservoir; Water quality assessment; Water quality index; Chlorophyll-a; Water quality assessment; Water quality index; Three Gorges Reservoir; Machine learning
• ISSN: 0048-9697; 1879-1026
• DOI: 10.1016/j.scitotenv.2023.169819

The degradation of water quality induced by the construction of large-scale hydraulic projects is one of the primary public concerns; however, it is rarely addressed with long-term field observation data. Here, we reported the long-term (2003–2021) trends, seasonal patterns, and overall condition of water quality of the Three Gorges Reservoir (TGR) with an enhanced water quality index (WQI). Specifically, to emphasize the importance of the biological role in water quality assessment, chlorophyll-a (Chla) was incorporated into WQI, and then a novel workflow using machine learning approach based on Random Forest (RF) model was constructed to develop a minimal water quality index (WQImin). The enhanced WQI indicated an overall “good” water quality condition, exhibiting a gradually improving trend subsequent to the reservoir impoundment in 2003. Meanwhile, the assessment revealed that the water quality has discernible seasonal patterns, characterized by poorer conditions in the spring and summer seasons. Furthermore, the RF model identified Chla, dissolved oxygen (DO), ammonium nitrogen (NH4-N), water temperature (WT), pH, and total nitrogen (TN) as key parameters for the WQImin, with Chla emerging as the most important factor in determining WQImin in our study. Moreover, weighted WQImin models exhibited improved performance in estimating WQI. Our study emphasizes the importance of biological parameters in water quality assessment, and introduces a systematic workflow to facilitate the development of WQImin for accurate and cost-efficient water quality assessment. Furthermore, our study makes a substantial contribution to the advancement of knowledge regarding long-term trends and seasonal patterns in water quality of large reservoirs, which provides a foundational basis for guiding water quality management practices for reservoirs worldwide. [Display omitted] •Long-term (2003–2021) evolution trend of water quality in the TGR was assessed by an enhanced WQI.•The water quality is maintained at “good” condition with a gradually improving trend during study period.•A novel workflow based on machine learning was proposed to develop cost-efficient WQImin models.•Chla is a key biological parameter in WQI-based water quality assessment.•WQImin consisting of Chla, DO, NH4-N, WT, pH, and TN showed excellent performance in water quality estimation.