SPATIO-TEMPORAL DYNAMICS OF WATER QUALITY IN LONGGAN LAKE BASED ON PRINCIPLE COMPONENT ANALYSIS (PCA) AND SELF-ORGANIZING MAPPING NEURAL NETWORK (SOM) MODELLING

In order to evaluate the spatiotemporal dynamics of water quality during the transformation stage of lake fishery model, we selected Longgan Lake, a typical lake in the middle and lower reaches of the Yangtze River, as our research site. From 2017 to 2018, seasonal water quality monitoring were conducted on the Huangmei waters and Susong waters of the lake. Principle component analysis (PCA) and self-organizing mapping neural network (SOM) modelling were used to analyze the spatiotemporal changes of physical and chemical parameters of water body. The eutrophication status of water body was evaluated by the method of comprehensive trophic level index (TLI). PCA results indicated that the water quality of Susong and Huangmei waters in Longgan Lake had little difference, while the seasonal dynamic was obvious. The average concentration of ammonia nitrogen in the whole lake was the highest (0.64 mg/L) in summer. The average concentrations of total nitrogen in summer and in winter were 2.30 mg/L and 1.04 mg/L respectively. The average concentrations of chlorophyll a in summer and autumn were 95.28 μg/L and 28.30 μg/L respectively. The pH value was the highest (9.27) in summer, while the average concentration of total phosphorus was highest in winter with the value of 0.22 mg/L. The TLI index showed that Longgan Lake had a mild eutrophic level in autumn and a moderate eutrophic state in the other three seasons. The results of SOM model clearly and intuitively reflected the temporal and spatial distribution of water quality in Longgan Lake. Management measures such as eliminating purse seine and prohibiting fishing can help restore lake fishery environment and fishery resources. It is suggested that a long-term follow-up investigation and assessment should be carried out on the lake ecosystem after the transformation of fishery model.