Abstract: This study examines the impact of direct assimilation of FY-3D satellite MWHS II microwave humidity sounder data on the prediction of extreme rainfall, using the July 31, 2023, heavy rainstorm in Beijing as a case study. Comparative experiments were conducted before and after data assimilation, and the WRF numerical prediction model was applied to analyze the effects across multiple scales and variables. The results show that the assimilation of MWHS II data significantly improved the simulation of extreme rainfall. It successfully captured the maximum rainfall center, exceeding 550 mm, and provided a more accurate simulation of rainfall distribution. The study also highlights the effect of assimilation on large-scale systems. It improved the large-scale environmental field, creating conditions that favored the extreme rainfall event. Key improvements included a strengthened temperature gradient in critical areas, optimized water vapor distribution, especially over the eastern sea, and an increased north-south pressure gradient. Together, these factors maintained a stable large-scale background that supported precipitation. On a smaller scale, the impact on convective systems was even more noticeable. Over the Beijing area, the vertical vorticity structure was optimized, with enhanced negative vorticity in the mid-to-upper atmosphere and increased positive vorticity in the lower levels, which promoted upward motion. The atmosphere became more unstable, with increased relative humidity in the lower levels, decreased humidity in the mid-levels, and a steeper vertical temperature gradient. These factors contributed to the triggering and maintenance of strong convection. Additionally, the microphysical processes were improved. More snow and graupel particles formed in the mid-to-upper layers, and the conversion of cloud water to rainwater accelerated in the lower levels, enhancing the overall precipitation efficiency. These effects were most prominent during the first 36 hours of the simulation, emphasizing the critical role of data assimilation during the early and developing stages of precipitation.