Abstract: The integrated global radiosonde archive (IGRA) is developed by the American national climatic data center (NCDC), and the dataset includes the multi-level, multi-daily radiosonde data elements of 1538 worldwide sounding stations from 1938 till now. The IGRA has the highest space-time density, and is the most complete and high-quality sounding dataset in the world. The IGRA dataset and its quality control were introduced and its data quality was analyzed. The results indicate that although this dataset has data quality control, it still has some problems. For example, the inconsistent rate of data from the IGRA and NMIC (Chinese national meteorological information center) datasets exceeds 70% for the sounding temperature data in China before 1973. The improper wind speed data from the IGRA are mainly from 1975 to 1978 and in 1981 and its rate accounts for 50.74 % among the wrong wind data in the above five years, and most wrong wind data concentrates in stations of the south American, the middle and south Africa and Russia. Most radiosonde data in the world are transported by the global communication system after 1980, so the quality and integrity of the radiosonde data from the IGRA after 1980 are relatively high. This paper can offer a reference to dataset users.

Key words:  IGRA dataset, Globe, Radiosonde data, Data quality