The radiosonde data is the most basic routine observation data in meteorological data assimilation and plays an important role in the effective improvement of assimilation analysis and forecast.Due to the low spatial resolution and uneven distribution of the existing radiosonde stations, only two observations per day and the small quantity, the accurate reproduction ability of their analysis field to the small and medium-sized atmospheric state is limited.After the L-band radar-digital radiosonde in our country has been updated, the radiosonde observation could obtain the atmospheric profile data 4 times a day with second-level and minute-level vertical resolution.In this paper, the WRF mesoscale numerical model is used to analyzethe sensitivity of intensive temporal radiosonde observationdata to the assimilation analysis and forecast quality through the effective assimilation of 06 UTC intensive radiosonde data and 12 UTC conventional radiosonde data.The results show that:the quality of afternoon rainstorm forecast by assimilating 06 UTC intensive radiosonde data is better than that of 12 UTC conventional radiosonde data.To be specific, the forecast skill of heavy rain and rainstorm by assimilating 06 UTC intensive radiosonde data are more accurate than by assimilating 12 UTC conventional radiosonde data.The root mean square error of prediction fields such as geopotential height, temperature and wind field is more effective at higher levels.The assimilation of the intensive radiosonde data at 06 UTC contributes more to the forecast quality of upper-level jet stream and lower-level water vapor flux divergence.Batch tests further confirm that assimilating 06 UTC intensive radiosonde data is of positive significance for both analysis fields and numerical forecasting.