3.3 Cluster Analysis and ANOVA Analysis under the time of water quality parameters
Cluster Analysis (CA) divides the sampling time into clusters depending on similar characteristics of the water quality indicators, which in turn makes an essential contribution to the overall analysis of the water quality later on (Li et al., 2018; Singh et al., 2004; Varol, 2020). In this study, Temporal CA was plotted in a tree diagram based on the changes in 10 metrics for the three reservoirs from 2019 to 2021 (Fig. S3). and clustered the 12 months at (Dlink / Dmax) × 100 < 15. Interestingly, all three reservoirs were categorized into three statistically significant clusters, and the three clusters corresponded closely to the stabilized water level period, rainy season high flow period, and winter low flow period in Chuzhou City, respectively. These two phenomena provide strong evidence of the similarity in the variation of water quality indicators in the three reservoirs. ANOVA results confirmed significant differences between clusters. For example, all water quality parameters except SD, BOD, NH3-N, and TP in the Shahe Reservoir showed significant differences (P < 0.05) between clusters (Table 3), for Huanglishu Reservoir, WT, pH, and DO were excluded (Table 4); and compared to the Shahe Reservoir, the Chengxi Reservoir contains more NH3-N and TP (Table 5).