Based on the temperature data collected by 7 stations in Qingdao and the 2meter ECMWF fine grid forecast data form 2013 to 2015, the forecast effectiveness of the EC model is verified in different seasons and synoptic systems. The results show that the forecast errors in coastal regions are greater than those in the inland, and the effectiveness becomes worse with the increase of lead time. Maximum temperature forecast has negative errors except the Jiaozhou station, and minimum temperature forecast has positive errors for the Qingdao, Pingdu, Laixi stations, and negative errors for the rest. The maximum temperature forecasts in March and April, as well as August and September, are not so stable, while the minimum temperature forecasts in the summer half year are better than those in the winter half year. According to the regularity of the errors, the amended reference values of the 7 stations are given. After correction, the forecast effectiveness is improved obviously, 3% to 16% for maximum temperature, and 4% to 18% for minimum temperature. The forecast performance in different weather is also different: in rainstorm, severe convection, high temperature, warmer, temperature dropping weather, the maximum temperature forecast is lower, and the minimum temperature forecast is lower in winter fog weather; when there is radiation cooling, the minimum temperature forecasts are lower at the coastal stations in the south and higher at the inland stations in the north.