Nowcasting is a critical technology for disaster prevention and mitigation, and the accuracy of radar echo extrapolation directly impacts forecasting performance. In most deep learning-based models, accurately predicting heavy precipitation remains a challenging task. Focusing on the region of China, this study proposes an improved model based on residual and attention mechanisms—RA-UNet—for precipitation nowcasting with a lead time of 3 h. The model introduces the residual neural network (ResNet) and the convolutional block attention module (CBAM) to integrate multi-scale features into the U-Net encoder–decoder architecture, enhancing its ability to capture the spatiotemporal evolution of precipitation systems. Meanwhile, depthwise separable convolutions are employed to replace conventional convolutions, significantly improving computational efficiency while preserving model performance. To evaluate the model’s performance, experiments were conducted using 6 min resolution radar echo data from China in 2024, with comparisons made against the optical flow (OF) method and the U-Net model. The experimental results show that RA-UNet demonstrates significant advantages in 3 h forecasting: its mean absolute error (MAE) is reduced by approximately 7%, the false alarm rate (FAR) decreases by about 20%, and it outperforms the comparison models in metrics such as the critical success index (CSI) and structural similarity index (SSIM). Notably, RA-UNet effectively mitigates intensity degradation in long-term forecasts, successfully predicting the trend of >40 dBZ strong echo cores in two typical cases and significantly improving the premature dissipation problem of precipitation fields. This study provides a new approach to refined forecasting of complex precipitation systems, and future work will combine multi-source data fusion with physical constraint mechanisms to further enhance precipitation event prediction capabilities.

An investigation is undertaken to explore a sudden quasi-linear precipitation and gale event that transpired in the afternoon of 30 May 2024 over Beijing. It was situated at the southwestern periphery of a double-center low-vortex system, where a moisture-rich belt efficiently channeled abundant warm, humid air northward from the south. The interplay between dynamical lifting, convergent airflow-induced uplift, and the amplifying effects of the northern mountainous terrain’s topography creates favorable conditions that support the development and persistence of quasi-linear convective precipitation, accompanied by gale-force winds at the surface. The study also analyzes the impacts of five microphysics schemes (Lin, WSM6, Goddard, Morrison, and WDM6) employed in a weather research and forecasting (WRF) numerical model, with which the simulated rainfall and radar reflectivity are compared against ground-based rain gauge network and weather radar observations, respectively. Simulations with the five microphysics schemes demonstrate commendable skills in replicating the macroscopic quasi-linear pattern of the event. Among the schemes assessed, the WSM6 scheme exhibits its superior agreement with radar observations. The Morrison scheme demonstrates superior performance in predicting cumulative rainfall. Nevertheless, five microphysics schemes exhibit limitations in predicting the rainfall amount, the rainfall duration, and the rainfall area, with a discernible lag of approximately 30 min in predicting precipitation onset, indicating a tendency to forecast peak rainfall events slightly posterior to their true occurrence. Furthermore, substantial disparities emerge in the simulation of the vertical distribution of hydrometeors, underscoring the intricacies of microphysical processes.