Abstract:After the radiosonde system of radiosonde station in China was upgraded to the L band sounding system, the high resolution vertical second level or minute level sounding data including temperature, pressure, humidity, wind profiles at 120 national stations can be continuously and automatically obtained. In this study, the high resolution vertical second level sounding profiling data are used to analyze the characteristics of their horizontal and vertical structures and to apply to GRAPES 3DVAR Regional Variational Assimilation Prediction System so as to compare the difference between the high resolution vertical second/minute level sounding profiling data and the traditional ones applied on the assimilation and prediction. The experiments using the data of June and July in the year of 2011 show encouraging results. The second level sounding profiling data can lead to a uniform horizontal distribution with highly dense vertical structure. The analyses of height field at 850 hPa and 500 hPa layers are closer to NCEP analyses than those with the traditional sounding data, which shows the improved effect for the analysis fields. Meanwhile, there is a significant improvement on the wind analysis fields for 200 hPa and higher layers, which shows the application of high resolution data on the high layer analysis can lead to more accurate system simulation. Furthmore, due to the improvement of the initial fields for the model prediction, the performance score for precipitation forecast increased and the forecast deviation obviously becomes smaller. Conclusively, the use of high resolution vertical radiosonde data has a positive impact on the initial analysis and forecast improvements of numerical model, which can provide the necessary foundation for the further development and utilization of second level radiosonde data.