The dataset used in this study is consist with observation and model results. Lightning data is obtained from regional Beijing Broadband Lightning Network (BLNet) comprised by 16 stations, which could detect the total lightning including intra-cloud lightning and cloud-to-ground lightning with lightning information of location, polarity, type, height and detection error. The average detection efficiency for total flashes is 93.2%, and the identification efficiency of BLNET for CG flashes is 73.9%. Radar data is obtained from the operational radar network of five operational S-band or C-band Doppler radars around Beijing area with the horizontal resolution of 0.01˚ (latitude) × 0.01˚ (longitude). The version of WPSv4.1.2, WRFDAv4.1.2 and WRFv4.1.2 are used in this study. Lightning data assimilation scheme (LDA) was proposed based on the relationship between lightning and updraft. The lightning data used for assimilation is provided in the dataset. The dataset volume is 85M.