Abstract: In this paper, the key technical features of CMA-BJ v2.0, the Hourly Rapid Catch-up Cycling Assimilation and Forecast System, are introduced in detail. This system uses IAU (Incremental Analysis Update) as the initialization scheme, which effectively suppresses the initial noise accumulation problem. By fully considering the actual truncated time of all kinds of observations' arrival, the two coupling parts, e.g., the cycle analysis and forecast update implementing data with different cut-off times, run in turn within each hourly cycling to meet the high demands of nowcasting and short-term forecast services. By applying the dynamic forecast hybrid scheme to the assimilated background field, the dynamic constraint of large-scale fields from global models on the development of micro- and meso-scale thermal and dynamic fields in the regional model is realized, and the deformation of large-scale prediction fields caused by the accumulation of rapid update cycle prediction errors is effectively suppressed. In terms of data assimilation, the national-wide mosaic radar reflectivity is only assimilated at the stage of forecast update to avoid continuous accumulation of water vapor. The radar assimilation background field error variance and length scale strategy optimization effectively promote the application of radar assimilation effect. In addition, real-time assimilation of national wind profile radar observations is realized in the CMA-BJ v2.0 system.