Abstract:Correcting and interpreting the model’s temperature forecast is an important means of improve the accuracy rate of objective temperature forecast in the context of model’s system error and impact of terrain. In this paper bias correction and the quasi symmetrical mixed running training period MOS forecast systems are developed based on ECMWF fine resolution model products. With the different methods an optimal consensus forecast method of temperature is designed. The accuracy rates of different models and different objective methods of daily maximum and minimum temperatures are compared. The results show that bias running correction of daily maximum and minimum temperature forecast in 10 to 30 days can improve the ECMWF fine resolution model’s temperature forecast. Bias running correction can significantly improve the daily maximum and minimum temperature forecasts of the models in short range, especially for the central mountainous area and check stations of Shandong Province. Bias running correction of daily minimum temperature can give higher improvement of the model’s forecast than that of daily maximum temperature. The MOS system can improve the daily maximum and minimum temperature forecasts too, while the accuracy rates of ECMWF fine resolution model, bias correction and MOS temperature objective forecast are different for different regions and different seasons in Shandong Province. Running optimal consensus forecast method can give further improvement of daily maximum and minimum temperature forecasts by integrating the advantages of different objective methods.