In this paper, GPS occultation refractive index data are assimilated by threedimensional variation assimilation system of WRF model which is used to explore the improvement of its initial value through the numerical simulation of 4-6 July 2007 Meiyu. GPS occultation data adjustments to the initial data are mainly at lowlevel moisture field; For the temperature and height fields, adjustments are gradually increasing from bottom to top. The horizontal and vertical incremental analysis is able to explore the effects of assimilation on improving the forecast， as well as learn more about the spatial structure of the system as a way to explore the dynamic mechanism of its occurrence and development. Assimilated results show that adding GPS occultation data to precipitation forecasting is the most obvious because of the initial lowlevel moisture field adjustment. The negative divergence and positive vorticity increments are well corresponding to precipitation zones. And it enhances the dynamic conditions of precipitation.