三种全球预报产品中国区近地面气温短期预报效果检验

全球气象预报产品是扩散模式、空气质量模式的重要基础资料和前提条件，其误差直接影响模拟结果的准确度。为考察不同气象预报产品的误差，选取2016年6月至2017年5月GFS、ECMWF、T639三种全球气象预报产品，利用中国2100个地面观测站数据，对预报产品中近地面气温进行了对比，并分析了其在不同季节、不同区域的特征。结果表明：在中国区域三种气象产品气温预报存在偏低预报的趋势，其均方根误差的年平均值为2.60—3.52℃，相关系数的年平均值为0.89—0.92，平均绝对误差的年平均值为1.87—2.67℃。整体而言，EC表现最佳，其余依次为GFS、T639。气温预报误差存在季节变化特征，三种产品均方根误差与平均绝对误差均表现为夏秋季优于春冬季，相关系数表现为秋冬季优于春夏季。气温预报误差存在明显的地域差异，三种气象预报产品的气温误差空间分布特征较为相似，在中国华东地区误差值表现较低，在西南地区误差较高。同时，其误差水平在中国沿海地区表现较低，在地形复杂地区表现较高。

Verification of the short-term forecast of near-surface temperature using different global forecast products in China

The near-surface temperature has an important influence on the diffusion and transfer process of airborne radioactive materials.Global weather forecast products are important basic data and initial conditions for the diffusion model and air quality model.Their errors directly affect the accuracy of simulation results.In order to investigate the errors of different products, three kinds of global weather forecast products, the GFS (Global Forecast System), ECMWF (European Medium-Range Weather Forecast Center), and T639 were selected from June 2016 to May 2017.Using the observational data from 2100 ground stations in China, the near-surface temperature of the forecast products was compared and analyzed, and their performance characteristics in different seasons and regions were investigated.The results show that the temperature forecast of the three meteorological products in China tends to be lower.The annual mean value of RMSE (Root mean square error) is between 2.60 ℃ and 3.52 ℃.The annual mean value of the correlation coefficient is between 0.89 ℃ and 0.92 ℃.The annual mean value of the average absolute error is between 1.87 ℃ and 2.67 ℃.Overall, the performance of ECMWF is the best, followed by the GFS and T639.The temperature prediction error is different at different seasons.Based on the RMSE and mean absolute error of the three products, it indicates that the performances in summer and autumn are better.While based on the correlation coefficient, it shows that the performances in autumn and winter are better.There are obvious regional differences in temperature forecast errors.The spatial distribution characteristics in the temperature forecast errors from the three weather forecast products are similar.The error values are lower in the east of China and higher in the southwest of China.At the same time, the error is lower in the coastal areas and higher in the inner land areas.