Cloudiness and precipitation are important output parameters in numerical weather prediction (NWP) models, both for their own sake and because they strongly affect other parameters, e.g., surface temperature. The spin-up problem is an important reason resulting in low accuracy of forecasts during the early prediction stage (0–6 h), so it’s necessary to introduce cloud-related information to eliminate or weaken this problem. For typhoon prediction, microwave satellite data is crucial. Profiles of cloud microphysical parameters can be retrieved from the microwave imager using certain inversion technology. The microwave imager onboard Fengyun-3B (FY-3B MWRI), the microwave imager onboard Tropical Rainfall Measuring Mission (TRMM TMI), and the Advanced Microwave Scanning Radiometer (AMSR-E) onboard AQUA are selected to do this research. Experiments of initialization of the cloud microphysical information in Global and Regional Assimilation and Prediction System (GRAPES) during typhoon MA-ON activity are carried out to investigate their impact on forecasts. The results indicate that prediction of hydrometer parameters and surface rain rate can be faster through initialization of cloud information derived from satellite microwave observations in GRAPES model and there is positive contribution during the first 6 hours of the model integration. Retrievals from MWRI, TMI and AMSR-E have a good consistency, and fusion data with three kinds of retrievals shows a more positive impact.