Abstract: To improve the accuracy of model initial field, a preliminary analysis of the impact of initial condition adjustment using satellite-derived humidity on the precipitation forecasting is carried out. Taking the wet-adiabatic equation as integral equation, the humidity field on each standard isobaric surface derived from satellite T_BB-data is used to improve the model initial conditions. An inter-comparison study is carried on between the satellite-derived humidity and T106 objective analyzed humidity. It is found that there is a big difference between the model-outputs driven by two different humidity fields. The simulation results driven by satellite-derived humidity depict the upper air humidity distribution above the precipitation region very well. Thus, a data assimilation technique based on the meso-scale model MM5 is performed to assimilate the conventional observation data and the high-resolution satellite-derived data. A rainstorm case is simulated in order to compare the precipitation forecast capability with different initial fields. The experiments show that when the satellite-derived humidity is assimilated into the initial condition, both the intensity and the area of simulated precipitation are more close to the measured precipitation as compared with those only using the conventional observations.