Based on conventional meteorological observation data, ERA5 reanalysis data and numerical model data, the mesoscale characteristics, circulation background, physical quantity environment and predictability in different stages of typhoon Chaba rainstorm at northern Hubei are analyzed. The results indicate that: (1) This process occurred under the circulation background of the typhoon moving northward and the combination of the westerly trough. The 700-925 hPa strong southerly jet on the east side of typhoon Chaba, combined with the upper-level typhoon circulation, provided the dynamic lifting conditions and water vapour supply conditions for the heavy precipitation. According to the influence of the system, this process could be divided into two precipitation stages: spiral cloud belt and low pressure main body. Both stages had obvious characteristics of convective warm cloud precipitation, and the hourly rain was strong. However, the precipitation in the spiral cloud belt was short in duration and dispersed in scope, while the precipitation in the low pressure main body was long in duration, wide in scope and large in cumulative rainfall. (2) The primary factor contributing to the disparity in rain mass evolution between the two stages lay in the fact that the spiral cloud belt stage exhibited robust convective instability and potential energy, while vertical motion and water vapour convergence at mid-to-low levels were relatively weak. Consequently, convective triggering was more dispersed and short-lived, primarily driven by cold outflows generated by ambient wind fields and cold precipitation areas. Conversely, during the main stage of the low pressure main body, there was also strong convective instability. The ascending movement, water vapour transport and convergence formed by the convergence of the lower level and the divergence of the upper level were significantly enhanced compared with the spiral cloud belt stage. At the same time, the cold air from the north invaded into the energy front, providing favourable thermal, dynamic and water vapour conditions for a large range of strong convection for a long time. Additionally, topographic blocking effects induced backward propagation of mesoscale convective systems (MCS), which together with synoptic-scale systems gave rise to a “train effect”. (3) In this process, the short-term predictability of heavy rain was high, but the predictability of precipitation extremes and strong central falling areas was low. In comparison, the global model had more advantages for the precipitation forecast of the low pressure main body stage, among which ECMWF yielded the best results, while the mesoscale model had more advantages for the precipitation forecast of the spiral cloud belt stage.