Abstract: On the basis of further improvement and development of Variational Doppler Radar Analysis System (VDRAS), a rapid-updating 4-D variational analysis system focusing on convective-scale numerical simulation and aiming at nowcasting convective storm has been preliminarily set up and tuned. The system is based on rapid-updating 4-D variational assimilation (RR4DVar) techniques of multi-Doppler-radar observations, a 3-D cloud-scale numerical model with simplified microphysics scheme which includes rainwater evaporation cooling and precipitation sedimentation processes, and an adjoint model. The system can rapidly get low-level 3-D analysis fields including convective-scale dynamical, thermo-dynamical and microphysical structures with 12-18-min updating cycles by assimilating both reflectivity and radial velocity observations from 6 CINRAD Doppler radars in Beijing-Tianjin-Hebei region using the RR4DVar scheme. It also integrates 5-min observations from regional auto weather stations (AWS) and forecast results from a meso-scale numerical model. Allowing for a strong convective storm case occurred in the region on 22 July 2009, simulated results from a series of sensitivity experiments including control, full-troposphere and full-microphysics, meso-scale background, and radar data assimilation are analyzed. These results are also compared and evaluated using intensive local observations from four wind profiler radars, two microwave radiometers, and two boundary layer towers. Some key factors for the system to produce appropriate analysis fields are illuminated. The system using low-level settings with the simplified microphysics scheme has comparable skill with full-troposphere settings and full-microphysics scheme. In the system, most significant RR4DVar assimilation of radar observations can be obtained using two or three scanning volumes from each radar within an assimilation window. As an effective supplement to radar observations on the ground, the AWS data is also very important on the RR4DVar assimilation of radar observations and simulations of dynamical and thermo-dynamical structures at several lower model levels. The meso-scale background and dynamical constraint for the RR4DVar assimilation of radar observations are sensitive to convective-scale simulation in both cold and warm start updating cycles. Results also indicate the system can produce robust pre-storm environment features including low-level inflows, vertical wind shear, low-level small-scale convergence, updraft and warm tongues. On the other hand, storm-associated convective-scale structures including cold pools and outflows can also be reasonably analyzed by the system.