A chlorophyll-a prediction model for Meiliang bay of Taihu based on Support Vector Machine

A chlorophyll-a prediction model for the Sanhaobiao monitoring site is established by using Support Vector Machine(SVM),based on the measured data of the Meiliang bay from April 2010 to December 2011.The total solar radiation,comprehensive extinction coefficient,water temperature,total inorganic nitrogen,pH and the current chlorophyll-a are chosen as input variables,and the chlorophyll-a in 7 days is selected as the output variable.The comparisons between the simulations and the observations indicate that the model could precisely predict the dynamic changes of chlorophyll-a 1 week later.Sensitivity analysis for all input variables reveals that the current chlorophyll-a is the most important factor to the prediction results,followed by pH,total solar radiation,comprehensive extinction coefficient,water temperature and total inorganic nitrogen.