The graded precipitation and rainy/fine forecasting experiments are conducted by means of several consensus methods such as majority voting, weighted scoring, and multivariable regression on precipitation caused by weather systems from west in Chengde. The results indicate that when there is relative high individual TS’s and slightly varied total forecasting performance, the majority voting method can improve the forecasting accuracy greatly; when the accuracies of various numerical forecasting methods for graded precipitation forecasting varied greatly, the weighted scoring method performed better; when there are enough samples, the refined repression method is more feasible for rainy/fine forecasting and graded precipitation forecast.