Abstract:Radiosonde observations and surface observations, which are obviously different in spatio temporal resolutions, serve as the most commonly used and vital data sources in weather analysis and numerical prediction. Further studies are required on how to use the surface observation data with high spatio temporal resolution to supplement deficiencies and shortages in radiosonde observations in some areas, especially on the effect of using mountain station data on the circulation of low level troposphere and on the indicative significance to the refinement of wind field analysis of numerical weather forecast. By analyzing various methods used in data assimilation of mountain station observations, this paper investigates the contribution of the supplementary data of radiosonde observation to the data assimilation. As a test case, the surface data and radiosonde observation at 00 UTC 29 June 2013 are adopted for data assimilation, which takes the surface observations of mountain station as radiosonde data and surface data in assimilation analysis. 〖JP2〗The results indicate that when the surface observation of mountain stations is used as radiosonde data, they have positive effects on the 850 hPa and 925 hPa wind data assimilation analysis. The comparison of using the surface observation as radiosonde data and as surface data shows no apparent difference in 24 h forecasting test on the height and wind fields at 850 hPa and 700 hPa levels. However, the 24 h precipitation forecast results show a weak positive contribution to rainfall intensity and location prediction when using the surface observation of mountain stations as radiosonde data. Continuous precipitation forecast tests prove that in the southwest areas where more mountain stations are located, the precipitation forecasting scores for light rain, moderate rain and heavy rain with station observations used as radiosonde data are better than those with station observations as surface data, and their forecast bias are also smaller. But in the other areas of China, no obvious difference is found in precipitation tests by using mountain data as the radiosonde data or the surface data.