[Construction and Application Optimization of the Chl-a Forecast Model ARIMA for Lake Taihu].

As an important indicator of phytoplankton biomass in lakes, the chlorophyll-a (Chl-a) concentration reflects the abundance and variation of phytoplankton in the water. Based on the monthly monitoring data of Chl-a and environmental factors in Lake Taihu from December 1999 to August 2019, key environmental factors related to Chl-a and their relationships were found using the principal component analysis (PCA) method. A multiple linear stepwise regression model and an auto-regressive integrated moving average (ARIMA) model were developed to predict the monthly Chl-a concentrations. The results showed that the Chl-a concentrations in Lake Taihu exhibited clear seasonal change characteristics and an overall trend of a gradual increase. The changes in total phosphorus (TP), the permanganate index, monthly average temperature (MAT), and monthly rainfall (MR) matched the Chl-a concentrations relatively well, whereas the changes in total nitrogen (TN) and ammonium nitrogen (NH-N) lagged significantly. The PCA results showed that the increased phytoplankton biomass and consequent algae outbreaks in Lake Taihu were not limited to the effect of a single factor such as TN or TP, but were comprehensively affected by multiple factors such as TN, NH-N, TP, the permanganate index, MR, and MAT. Through further validation, the ARIMA model of Chl-a concentrations was proved to be significantly better than the multiple linear stepwise regression model, especially when considering the key environmental factors as independent variables and optimizing their values. The established ARIMA (0,1,1) (0,1,1) model would be helpful for forecasting algae blooms in Lake Taihu and provide useful suggestions for water environmental management, such as water resources dispatch and regulation.