A positive cloud‐to‐ground (+CG) lightning flash containing a single stroke with a peak current of approximately +310 kA followed by a long continuing current triggered seven upward lightning flashes from tall structures. The flashes were observed on 4 June 2016 at the Tall Object Lightning Observatory in Guangzhou, Guangdong Province, China. The optical and electric field characteristics of these flashes were analyzed using synchronized two‐station data from two high‐speed video cameras, one total‐sky lightning channel imager, two lightning channel imagers, and two sets of slow and fast electric field measuring systems. Three upward flashes were initiated sequentially in the field of view of high‐speed video cameras. One of them was initiated approximately 0.35 ms after the return stroke of +CG flash from the Canton Tower, the tallest structure within a 12‐km radius of the +CG flash, while the other two upward flashes were initiated from two other, more distant tall objects, approximately 18 ms after the +CG flash stroke. The initiation of the latter two upward flashes could be caused by the combined effect of the return stroke of +CG flash, its associated continuing current, and K process in the cloud. Each of these three upward flashes contained multiple downward leader/upward return stroke sequences, with the first leader/return stroke sequence of the second and third flashes occurring only after the completion of the last leader/return stroke sequence of the preceding flash. The total number of strokes in the three upward flashes was 13, and they occurred over approximately 1.5 s.

Needle structure is a recently discovered lightning phenomenon, first observed by LOw Frequency ARray (Bandwidth: 30-80 MHz) radio telescope having very high spatial and temporal resolutions, which extends sideways from the positive leader channels of intracloud (IC) and negative cloud-to-ground (CG) flashes (Hare et al., 2019). Furthermore, Hare et al. (2019) conjectured that the needle-like discharge was caused by an electric field reversal, which in turn was due to positive leaders transiently disconnecting from their negative counterparts. The analysis and discussion of the characteristics and possible mechanism of needles could lead to a better understanding of the processes involved in the transverse (radial), as opposed to longitudinal motion of charge in lightning channels. The radial motion of charge during the return-stroke process was previously studied by Rakov (2006, 2009).In the following, we will briefly review the characteristics of needles reported from observations with VHF interferometers and high-speed framing cameras. Needles initiating from the main positive leader channel of an IC lightning were imaged by a short-baseline VHF interferometer with 200 MHz bandwidth (Pu & Cummer, 2019), and it was found that these needles moved forward (away from the positive leader channel) continuously with density decreasing backwards. Hare et al. (2021) showed detailed characteristics of needle propagation and flickering and observed that recoil leaders quench needle activity. Saba et al. ( 2020) observed needles appearing on upward positive leaders, and their high-speed video records showed that needles were formed at failed leader branches. It is possible that needles are a kind of common (not rare) processes occurring in lightning discharges.

High‐speed video and electric field change data have been used to examine the initiation and propagation of 21 recoil leaders, 7 of which evolved into dart (or dart‐stepped) leaders (DLs) initiating return strokes and 14 were attempted leaders (ALs), in a Canton‐Tower upward flash. Three DLs and two ALs clearly exhibited bidirectional extension. Each DL was preceded by one or more ALs and initiated near the extremity of the positive end of the preceding AL. The positive end of each bidirectional DL generally appeared to be inactive (stationary) or intermittently propagated along the positive part of the preceding AL channel and extended into the virgin air. A sequence of two floating channel segments was formed ahead of the approaching positive end of one DL, causing its abrupt elongation.

Using high‐speed video cameras operating with framing rates of 20 and 525 kfps, we imaged the attachment process of a natural negative cloud‐to‐ground flash, occurring at a distance of 490 m. Nine upward leaders were observed. A total of 12 space stems/leaders in 47 steps of the downward negative stepped leader were captured. The two‐dimensional length of them was between 2.0 and 5.9 m, with an average of 3.0 m. The average interstep interval, step length, and two‐dimensional speed of the downward negative leader and that of upward positive leader were statistically analyzed. The last step of the downward negative leader making contact with the upward connecting leader was recorded. The two‐dimensional length of the final imaged gap between the tips of opposite‐polarity leaders was estimated to be about 13 m.

High‐speed video and electric field change data were used to analyze the initiation and propagation of four predominantly vertical bidirectional leaders making connection to a predominantly horizontal channel previously formed aloft. The four bidirectional leaders sequentially developed along the same path and served to form a positive branch of the horizontal in‐cloud channel, which became a downward positive leader producing a 135‐kA positive cloud‐to‐ground (+CG) return stroke. The positive (lower) end of each bidirectional leader elongated abruptly at the time of connection of the negative (upper) end to the pre‐existing channel aloft. Thirty‐six negative streamer‐like filaments (resembling recently reported “needles”) extended sideways over ∼110 to 740 m from the pre‐existing horizontal channel at speeds of ∼0.5 to 1.9 × 107 m/s, in response to the injection of negative charge associated with the +CG.