Abstract: After making landfall in Shanghai, Typhoon Bebinca (2413) gradually moved northwestward. It entered Anhui Province from Jiangsu Province at severe tropical storm intensity and induced a persistent heavy rainfall process in Anhui. This process can be divided into two stages: in the first stage, widespread heavy precipitation occurred across Anhui Province. in the second stage, after the typhoon weakened to a tropical depression, localized extreme heavy precipitation was triggered in northern Anhui. Based on multi-source observational data, ERA5 reanalysis data, and numerical model forecast products, this study compares the environmental conditions, mesoscale convective systems, and precipitation microphysical characteristics of the two rainfall stages, and explores the causes of the significant forecast bias of the model in the second stage. The main conclusions are as follows. (1) The heavy precipitation in both stages was dominated by high concentrations of small particles, indicative of tropical maritime precipitation. A substantial number of small particles grew through collision-coalescence during their descent, leading to the presence of larger particles in surface rainfall. (2) In the first stage, strong upward motion and water vapor convergence were present in the southeastern quadrant of the typhoon. As the typhoon moved northwestward, it triggered widespread heavy precipitation in Anhui. In the second stage, after the typhoon weakened to a tropical depression, the upward motion and water vapor convergence zones shifted to northern Anhui but were weaker than in the first stage, and the mid-to-low-level upward motion zone tilted northward with height. (3) Influenced by spiral rainbands, banded echoes exhibiting the train effect formed over northern Anhui. The sustained, efficient warm-cloud precipitation process led to localized extreme heavy precipitation in this region during the second stage. (4) The ECMWF model forecast showed a westward bias in the typhoon center position, with stronger low-level northerly winds to the north of the typhoon, resulting in a westward bias in the forecast heavy precipitation area for the second stage. Once mesoscale convection was triggered, the latent heat release further amplified the model forecast bias.