Abstract: In this paper, the mesoscale numerical weather prediction model WRF3.2 is used to study the influence of different assimilation time intervals on the heavy rains simulation, which occured on July 3, 2011 in Chengdu.The results show the assimilation of radar data can make the initial field distribution of water vapor field, horizontal wind field and vertical velocity field more closer to the fact.Except temperature field, the frequency of assimilation is higher, which can bring about greater impact on the initial field, greater incremental value of the horizontal wind field, and The more information on the medium and small scale are present, which can be more closer to the real atmosphere.After many times assimilations, low level wind field changes have been delivered to the high level, which cause adjustment of the flow field in high level.Radar reflectivity data assimilation is effective, which can adjust the distribution of water vapor in atmospheric.