This study evaluates how clouds evolve into thunderstorms in terms of the turbulence characteristics producing the first flash. Observations of 57 (39) isolated thunderstorm (non‐thunderstorm) cells during 2016/2017 in Southern China are provided by an S‐band polarimetric radar and three independent lightning location systems. The vertical turbulence characteristics of clouds associated with thunderstorms are obviously different from non‐thunderstorms. For thunderstorms, the maximum of the eddy dissipation rate (mean value in each height layer) in the entire height is 0.19 m² s⁻³, which occurs at the first flash stage, and the achievable height of turbulence exceeds the −30°C layer. For non‐thunderstorms, however, the maximum is 0.12 m² s⁻³, and the achievable turbulence height hardly exceeds the −10°C layer. Additionally, the turbulence intensities of the locations where the initial discharge pulse events of the first flashes occur are weak. These turbulence characteristics are useful for lightning nowcasting.

Determining whether a cloud will evolve into a thunderstorm is beneficial for understanding thunderstorm formation and also important for ensuring the safety of society. However, a clear understanding of the microphysics of clouds in terms of the occurrence of lightning activity has not been attained. Vast field observations and laboratory experiments indicate that graupel, which is rimed ice, is a vital hydrometeor for lightning generation and is the foundation of riming electrification. In this study, polarimetric radar and lightning observations are used to compare the ice microphysics associated with graupel between 57 isolated thunderstorms and 39 isolated non-thunderstorms, and the differences in radar parameters are quantified. Our results for the occurrence of lightning activity in clouds revealed the following results: (1) the maximum difference in graupel volume at the −10 °C isotherm height between thunderstorms and non-thunderstorms reached approximately 7.6 km³; (2) the graupel particles approached spherical shapes, with a mean differential reflectivity (Z_DR) value of 0.3 dB, which likely indicated that heavily rimed graupel was present; (3) the median values of horizontal reflectivity (Z_H) or Z_DR at positions where the source initiation and channel of the first lightning flashes were nearly 31 dBZ or 0 dB; and (4) 98.2 % of the thunderstorms were equipped with a Z_DR column, and the mean depth was ∼ 2.5 km. Our study deepens our understanding of lighting physics and thunderstorm formation.

Polarimetric structures detected by radar can characterize cloud microphysics and dynamics. Many studies have indicated that differential reflectivity (Z_DR) and specific differential phase (K_DP) columns, which serve as proxies for updraught strength, are related to lightning activity; moreover, the quantities of ice and supercooled liquid water strongly influence the occurrence of lightning flashes via non-inductive charging. However, the sequence or interactions among these factors with dynamics and microphysics from the perspective of the cloud life cycle are uncertain. Here, we improve the “3D mapping columns” method to identify and quantify the $Z_{\mathrm{DR}}/K_{\mathrm{DP}}$ columns, which is based on Cartesian grid datasets; this method is sensitive in the early phase of cloud formation. Our study bridges the polarimetric structure and lightning activity within 15 isolated thunderstorms during the cloud life cycle. The results indicate that microphysical variations in supercooled liquid water and graupel yield better correlation coefficients for the lightning activity prediction at short warning times (e.g. 6 min) than dynamical variations in the Z_DR column volume do; however, the trend of the Z_DR column volume implies good performance at longer warning times (e.g. 12 min). The K_DP column is probably absent in the early phase of convection development; however, it will occur in the later stage with heavily cold cloud processes, replacing the Z_DR column to indicate updraughts within the reflectivity core when obvious graupels and hailstones occur. Our study improves the understanding of the polarimetric structure, which is related to dynamics and microphysics, and is also associated with lightning activity.

The determination of whether a cloud will evolve into a thunderstorm is beneficial for understanding thunderstorm formation and important for ensuring the safety of society. However, a clear understanding of the microphysics in clouds for the occurrence of lightning activity has not been attained. Vast field observations and laboratory experiments indicate that graupel, which is rimed ice, is a vital hydrometeor for lightning generation, and is the foundation of riming electrification. In this study, polarimetric radar and lightning observations are used to compare the ice microphysics associated with graupel between 57 isolated thunderstorms and 39 isolated non-thunderstorms, and the differences in radar parameters are quantified. Our results for the occurrence of lightning activity in clouds showed the following results: 1) the maximum difference in graupel volume on the −10 °C isotherm height between thunderstorms and non-thunderstorms reached approximately 7.6 km³; 2) the graupel particles approached spherical shapes with a mean Z_DR value of 0.3 dB, which likely indicated heavily rimed graupel was present; and 3) 98.2 % of thunderstorms were equipped with the Z_DR column, and the mean depth was ~2.5 km. Our study deepens our understanding of lighting physics and thunderstorm formation.

This study improved the first lightning flash (FLF) warning method of isolated thunderstorm cells in the warm season over South China. The S-band polarimetric radar and three independent lightning location systems provided 57 thunderstorm and 39 nonthunderstorm cells. The results indicated the median value of horizontal reflectivity below the melting layer when the cloud detected by the radar first was a good indicator for forecasting the FLF with the longest lead time (threshold value ≤ 15 dBZ in this study). The algorithm produced 0.96–1 probability of detection (POD), 0.32–0.38 false alarm ratio (FAR), 0.63–0.66 critical skill index (CSI), and 12–21-min average lead time. Methods of minimum echo (20 dBZ) top height extending to −10°C and the 35 dBZ at −10°C were more accurately for warning the FLF but with a shorter lead time; the former algorithm resulted in 1 POD, 0.08–0.29 FAR, 0.71–0.92 CSI, and 5–9-min average lead time. The latter algorithm resulted in 1 POD, 0–0.06 FAR, 0.94–1 CSI, and 3–5.5-min average lead time. The fluctuation of lead time depended on the FLF occurrence time. The differential reflectivity Z_DR column and the height of graupel were not as effective as the reflectivity information on warning the FLF in this study. The results are expected to improve the FLF forecasting in considering accuracy and lead time within warm-season thunderstorm cells over South China based on the radar. Significance Statement: It is difficult to forecast the first lightning flash (FLF) occurrence. While many methods are provided based on weather radars, the threshold values for hitting the start of lightning activity are dominated by mixed-phase ice microphysics, especially associated with graupel particles; therefore, to improve the lead time of the FLF nowcasting, usage of knowledge of warm-phase microphysics is possible. Fortunately, the latest discovery indicates the difference is recognizable in warm-phase microphysics during the first radar echo between thunderstorms and nonthunderstorms based on the weather radar. This study utilized the finding to construct a method for the FLF warning. The early lead time and accuracy are considered comprehensively in this method. It is important for the FLF warning.