Abstract:The prediction of water quality is the foundation of the analysis,control and treatment of water pollution.Based on the data of water quality indexes of COD and NH3-N from 2004-2011 at the Jishui water port of Qingpu District,Shanghai City in the Taihu Lake basin,the equal dimension and new information model was established for each index.The mean relative error of the NH3-N index did not meet the standand,so the Markov model was used to improve the predicted values.Then,the future water quality indexes were predicted by the two models.Results show that the equal dimension and new information Markov model effectively reduces the average error of the predicted values and enhances the accuracy of them.In the next few years,the two water quality indexes show a downward trend and tend to be flattened to a certain extent gradually.However,due to the long-term nature of the pollution of the Taihu Lake basin,the water environmental situation remains grim.So,reasonable and effective measures should be taken in a long period of time,such as reducing source pollution,making ecological dredging,accelerating the transformation of economic development mode in the basin and improving the management of the basin.The water quality of the Taihu Lake basin may be overall improved in the future.