Time Series Modeling on 1998-2006 Chlorophyll-a in Taihu Lake, China

Due to data unavailability, there is little research of Chlorophyll-a modeling on Taihu Lake, China. In this paper, time series modeling was used to analyze 1998-2006 monthly Chlorophyll-a data. First, exploratory analysis, trend analysis, correlation functions and spectral analysis were used to acquire a general picture of the data. Second after the data division (1998-2005 for model estimation and 2006 for comparison), SARIMA model was applied with AIC and BIC as good-fit criteria. Then further diagnostic checks were carried out to test the fitted model. Although the general trend is right, the forecasted value is smaller than the real value of 2006, especially in May, June and August. In conclusion, although SARIMA model has captured some part of characteristics, additional variables such as temperature may need to be considered.