A lightning flash striking a 325‐m tower was captured using a high‐speed camera operated at 380 kfps and the characteristics of breakthrough phase (BTP) were recorded in two consecutive frames. The fine time‐resolved images indicate that the common streamer zone (CSZ) formed when the bright tips of negative and positive leaders were ≥23 m apart from each other. During BTP, both the negative and positive leaders extended their channels with enhanced luminosity and the gap between them was shortened to 16 m in the frame prior to return stroke. A space leader‐like luminous segment was found within the route of leader connection. This route is inferred to have already been determined at the initial formation of CSZ, by the interconnected streamers of approaching leaders. The dominant growth of this route is interpreted as a new scenario for the conversion of a high‐impedance CSZ to a hot plasma channel during BTP.

We conducted a comprehensive observation of tower‐initiated upward lightning (UL) during the summer season from 2012 to 2020, using high‐speed video and the Beijing Broadband Lightning Network (BLNET). A total of 25 UL flashes were observed, and 21 of them were triggered by nearby positive cloud‐to‐ground (CG) lightning. It is interesting to note that eight (38%) of the parent +CG lightning flashes have more than one return stroke (RS). The distance between the UL and the grounding point of its parent +CG lightning was 4–38 km, where nine (43%) of the UL were triggered less than 2 ms after the positive RS, and 11 (52%) of them were triggered after a long time delay of 5–187 ms. Based on high‐speed video and mapping results, the short time delay was associated with the reactivation of the decayed established leader above the tower. The long time delay was associated with the negative leader extending to the region above the tower during the long continuing current. The positive E‐field change during RS or continuing current period favored the inception of an upward leader. Although both preliminary breakdown (PB) and RS in a +CG flash produced positive E‐field changes, an upward positive leader induced by PB aborted its sustained propagation after +RS.

Quickly identifying and classifying lightning waveforms is the foundation of lightning forecasting and early warning. In this paper, based on the electric field observation of the Beijing lightning location website of the Institute of Atmospheric Physics, Chinese Academy of Sciences, a recognition and classification method of pulse signal waveform based on Convolutional Neural Network(CNN) algorithm is designed and implemented. The CNN network model and its parameters were optimized from three aspects: dataset, model parameters, and network structure, achieving a recognition rate of over 90%. The effects of various optimization terms and their different optimization orders on the training time of the model were studied. The results indicate that the CNN algorithm is suitable for the classification and recognition of lightning electric field (LEF) waveforms. Optimization can significantly improve recognition rate. The optimization method of fitting idealized waveforms can reduce noise in the dataset and significantly improve recognition rate, indicating that noise has a significant impact on waveform recognition. Therefore, it is necessary to perform noise preprocessing before recognition. The optimization has a huge impact on training efficiency, increasing training time by about 51% after optimization, but the influence of optimization order on it can be ignored.