ABSTRACT
BP network has been adopted to forecast the highest point of nonlinear about chlorophyll (Zhu S.P et al, 2011). A prediction model combined with WT and Neutral Network has been developed and applied to bloom prediction in lakes of Beijing (Wu Q.M et al. 2010). The decision tree and piecewise nonlinear regression method of statistics has been applied to predict the variation trend of chlorophyll-a in the coastal zone of Holland (Chen Q & Mynett A.E. 2004). A hybrid model combined with ANN and GA has been developed to make prediction towards turbidity and chlorophyll-a that this model could make real-time forecast as well as mid-long term forecast (Iyan E.M et al. 2013). The method combined with machine and experience has been used to make a three-day short-term prediction towards environmental factors of Chesapeake Bay such as the temperature, chlorophyll-a, Nitrogen and DO (Brown C.W et al. 2013). A hybrid model combined with SVR and GA has been developed to make prediction towards blooms which has successfully confirmed the meaning of existence and growth of blooms influenced by biological and chemical variables in lakes (García Nieto P.J et al. 2013).
