The radiosonde observation is one of the most important data in the data assimilation system. The Russia radiosonde observation was adjusted in January 2015, and reduced some observations at 0000 UTC and 1200 UTC. In order to verify the effect on GRAPES model, the impact experiment was carried out. The observation was retrieved from National Meteorological Information Centre of China Meteorological Adminisration (NMIC/CMA) database. The GRAPES wind analysis of Russia Region became worse because of the Russia radiosonde adjustment, but the impact on the other regions was not so much. The height analysis became a little better in Russia and East Asian Regions, which may be related to the height observation quality. From the changes of anomaly correlation coefficient of the 500 hPa height forecast, the adjustment has positive effect on the forecast for East Asia.