Based on the weather radar and radiosonde data, a widely used thunderstorm identification and extrapolation algorithm is used in this paper to compare the characteristic difference of probability frequency and membership for thunderstorms with or without cloud-to-ground (CG) lightning activity, and a one-hour warning method is established on the principle of fuzzy logic by extracting the radar parameters. The analysis results show that the maximum radar reflectivities in the -25℃ to -10℃ layer and at the of echo top (ET) are the best factors for CG warning in Hubei Province. In particular, REF-15℃, REF-20℃ and REF-25℃ have the best performance, followed by REF-10℃. By contrast, vertically integrated liquid is less indicative of lightning activity. Thus, according to the different contributions of these factors, the unequal weight distribution is given, and the dynamic weight coefficient is used to finally achieve the location forecast of CG lightning in the future one hour. In addition, the 1 km grid point-to-point verification is performed by using the 1 h (6 min interval) nowcasts and in-site observations, and the results indicate that the probability of detection (POD) can reach more than 50% in 30 min, and the critical success index (CSI) is about 30%. However, POD and CSI decrease slowly with the increasing lead time of forecasts. Through the test of individual forecast case, it is found that this method is more suitable for the large-scale thunderstorm system, but relatively poor for local convective system. This 〖JP2〗finding illustrates that the fuzzy〖JP〗 logic CG lightning warning method based on radar parameters is reasonable 〖JP2〗and reliable, and can be used for automatic warning and decision-making service in Hubei Province.height