To provide radar product thresholds close to the actual for thunderstorm identifying, tracking and nowcasting systems, and to establish the algorithm of thunderstorm day discrimination based on radar observation, the artificial thunderstorm observation and radar data are used to construct the matching rate model, and the cumulative departure method is adopted to identify the matching rate of inflection point to determine the thresholds and the optimal radius of hearing thunders, and the echo area of the threshold is given. Taking the Changde meteorological station as an example, using this method, it is resulted that: (1) The threshold value of radar product is 38.5 dBz, and the optimal radius of hearing thunders reaches 25 km, which is larger than the radius obtained by comparing the lightning location data with the artificial thunderstorm day. The echo area threshold value is 73 km2, and the evaluation indicates that the threshold values are appropriate. (2) The seasonal variation of thunderstorms should be taken into account when using various thresholds and radar data to judge thunderstorm days. The performance is satisfactory in the Changde area from March to August. (3) When using this method, the matching rate model should be constructed based on the single radar or radar mosaic data samples of the longtime series selected according to the historical hearingthunder locations.