Abstract:

Based on temperature data from radiosonde dataset at 105 stations in China and three reanalysis datasets of NCEP/NCAR, ERA and JRA, the reliability of free atmospheric temperature in the northeast and northwest of China was analyzed by methods of a root mean square error analysis and a trend analysis and so on. The results indicate that in terms of climatological mean, the root mean square error of JRA datasets is lower than that of radiosonde dataset, and it is lower in winter and higher in the northwest of China than in the northeast of China. In terms of annual variance, atmospheric temperature of three reanalysis datasets are close to that of radiosonde dataset in low layer than in upper layer in the northeast of China, while temperature differences between three reanalysis datasets and radiosonde dataset are higher in the northwest of China than in the northeast of China. In the aspect of long-term variance, three reanalysis datasets have comparable ability in reproducing the trend in lower troposphere. Cooling trend in upper layer can be better identified by the NCEP reanalysis dataset in the northeast of China, while it is by the JRA reanalysis dataset in the northwest of China.

Key words: Radiosonde dataset, Reanalysis dataset, Climatological , mean, REOF (rotated empirical orthogonal function)