Abstract: The Kalman filter method is used to correct the errors of the predicted fields of the T63L16 model of the National Climate Center. The corrected fields are applied to a precipitation forecasting model which combines the qualitative analysis, based on characteristics of weather systems, with the quantitative analog, based on the analog discrete degree indexes, and the precipitation of 51 stations in Northeast China in the summer of 1996-1998 is made. The results show that its prediction performance is good.