Abstract: The auto nowcasting system for convective storm (BJ-ANC) is an expert system that has been improved and further developed to many key algorithms and modules other than they are just introduced from US. The skill of the system is focused on nowcasting convective storm in Beijing and its vicinity. Key improvements have been made to the algorithms including quality control of CINRAD radar data, diagnosis and analysis of local observations from radars, satellites, AWSs, rawinsondes and meso NWP results from a WRF based rapid update cycling model, identification, analysis and tracking of storm cells, grid tracking and extrapolation of radar echo, rapid updating assimilation of radar data, quantitative precipitation estimate (QPE) and quantitative precipitation forecasting (QPF), etc. The forecast factors produced by the above algorithms are closely related with initiation, evolution and decay of convective storm in general. These algorithms are integrated by a fuzzy logic algorithm under non dimensional modes with different weighting coefficients in the system and nowcasting products are generated. Based on many theoretical and experimental results, the relationship and conceptual model of boundary layer convergence line and initiation and evolution of storm have been imported into the system for nowcasting localized initiation and rapid evolution of storm. BJ-ANC system runs in real-time mode during 2008 summer time. The system performance for nowcasting initiation and rapid evolution of storm is improved after convergence line is adopted by interactive human entry function. Verification of storm cell extrapolation and 1 hour QPF in Beijing area and its vicinity during 2008 summer is performed using the B08FDP real time verification system. The results indicate extrapolation of storm cell in 1 hour is significant and the forecast deviation is small. The 1 hour QPF from the system is comparable with rainfall observation from AWSs, denoting the QPF products can be used for supporting nowcasting operation of convective rainstorm. The storm reflectivity nowcasts from the system have been demonstrated much more significant than those from persistence method. Storm case analyses indicate these products from the system are conductive for storm nowcasting operation and boundary layer convergence line is extremely helpful. In addition, retrieval of thermo dynamical fields can clearly indicate three dimensional structures of wind, convergence and temperature at low layer and further help forecasters make decisions on storm initiation and evolution nowcasts.