Analysis of Influencing Factors of Chlorophyll-a in Lake Taihu Based on Bayesian Network

Abstract

Global warming has aggravated the problem of lake eutrophication. As a typical large, eutrophic, shallow lake in China, the issue of cyanobacterial harmful algal blooms (cyanoHABs) was particularly prominent in Lake Taihu. We took Lake Taihu as the study area, using the meteorological (temperature, wind speed, rainfall, and sunshine hours), water quality (total nitrogen, total phosphorus, conductivity, pH, and chemical oxygen demand), and biological (chlorophyll-a in phytoplankton) monitoring data from 1992 to 2020. We built a simulation model of chlorophyll-a based on the Bayesian network model with continuous variables to study the chlorophyll-a level of Lake Taihu under different meteorological and water quality conditions. The 75th percentile of chlorophyll-a concentration was used as the threshold to judge the risk of cyanobacterial bloom. When the probability of chlorophyll-a concentration below this threshold was greater than 75%, it was regarded as "low risk" of cyanobacterial bloom outbreak. The results showed that the average level of "temperature wind ratio" (ratio of air temperature to wind speed) in spring was 6.67℃·s·m-1, and the probability of high chlorophyll-a was less than 75% when the total phosphorus concentration was less than 0.130 mg·L-1. The average "temperature wind ratio" level in summer was 10.52℃·s·m-1, and the probability of high chlorophyll-a was less than 75% when the total phosphorus concentration was less than 0.257 mg·L-1. The average level of total phosphorus concentration in autumn was 0.154 mg·L-1, and the probability of high chlorophyll-a was less than 75% when the "temperature wind ratio" was less than 6.30℃·s·m-1. Based on the above research, the chlorophyll-a model constructed by the Bayesian network model with continuous variables was further used to simulate the nutrient control objectives under different climate change backgrounds. In order to control chlorophyll-a in Lake Taihu at the:"low risk" level of cyanoHABs, the target concentration thresholds of total phosphorus needed to be controlled under the climate level background from 1992 to 2000, 2001 to 2010, and 2011 to 2020 were given. From 1992 to 2000, the threshold value of total phosphorus concentration was 0.135 mg·L-1 in spring, 0.174 mg·L-1 in summer, and 0.171 mg·L-1 in autumn. From 2001 to 2010, the threshold value of total phosphorus concentration was 0.115 mg·L-1 in spring, 0.164 mg·L-1 in summer, and 0.162 mg·L-1 in autumn. From 2011 to 2020, the threshold value of total phosphorus concentration was 0.059 mg·L-1 in spring, 0.145 mg·L-1 in summer, and 0.145 mg·L-1 in autumn. The results showed that the control of cyanoHABs in eutrophic lakes required more stringent nutrient control strategies with global warming. It provided a reference for preventing and controlling cyanoHABs and eutrophication in Lake Taihu. Previous studies have used multiple regression models, hydrodynamic numerical models, and other methods to predict chlorophyll-a concentrations or cyanobacterial blooms in lakes. However, there has been no study on the prediction of cyanoHABs in lakes based on the Bayesian network model with continuous variables and the "dynamic" evaluation of nutrient thresholds. Therefore, based on the seasonal meteorological, water quality, and biological monitoring data of Lake Taihu from 1992 to 2020, the chlorophyll-a model of Lake Taihu was constructed for the first time based on the Bayesian network model with continuous variables to simulate the chlorophyll-a concentration of Lake Taihu under different climate indicators and total phosphorus concentrations. The weight of its influencing factors was also analyzed, and the nutrient control objectives under different climate scenarios were "dynamically" evaluated.