By means of the sliding-window Chebyshev expansion technique, the objective forecast model (OFM) of heavy rainfall from May to June in the Jianyan station was developed in order to improve the capability of single-station rainfall forecasting. The results show that the objective forecast factors have clear meteorological meaning though simple and the advantages of high prediction accuracy and good stability; the characteristics of the weather patterns at three levels for various models exhibit the reasonable vertical collocations favorable to rainfall process development. The trial use in 2000 and 2002 indicates that the average accuracy of rainfall prediction of OFM is about 50% higher than that of the subjective, and there is no failure in 3 years, especially no miss and failure in 2003. It is indicated that the model is efficient and applicable for the improvement of single station rainfall prediction.