The location accuracy of the U.S. National Lightning Detection Network (NLDN) has been evaluated using as ground‐truth rocket‐triggered lightning data or video records but only at a few specific locations. In this study, by using the NLDN data for the events attributable to lightning strikes to towers, the location error of the NLDN across the entire contiguous United States was evaluated for the first time. We found that, on average, the NLDN median location error reduced from 198 to 84 m after the 2013 NLDN upgrade. The location error at the periphery of the network is significantly larger than that in its interior. In the coastal regions, there is directional location bias toward the water. Simulation results suggest that the bias is related to the lengthening of field waveform front due to electromagnetic wave propagation over lossy ground coupled with the asymmetrical sensor configuration relative to the strike point (lack of offshore sensors).

Using visible‐range and infrared (3–5 μm) high‐speed video cameras, we observed luminosity transients that reilluminated decayed branches of two close (2 to 4 km) negative stepped leaders in Florida. Leader branches were energized via stepping at their tips and, as a result, were most heated near their lower ends, with the hotter sections being connected via cooler sections to the trunk. In the modeling of lightning leaders, usually a single tip is considered. In contrast, in the present study, many (up to 30 per major branch) tips were active at the same time, forming a network‐like structure with a descending multitip “ionization front” whose transverse dimensions were of the order of hundreds of meters. The front exhibited alternating stepping, with each step necessarily generating a positive charge wave traveling from the leader tip up along the channel, like a mini return stroke. We inferred that the step‐related waves can cause luminosity transients in the remnants of decayed negative branches at higher altitudes. Such reactivated branches, in turn, may facilitate further leader stepping at lower altitudes, as first reported by Stolzenburg et al. (2015, https://doi.org/10.1002/2014JD022933). The reactivation process is likely to involve multiple steps, as evidenced by a large number of active tips (some tens per 50‐μs frame) and corresponding electric field pulses occurring at time intervals of 2 μs or less. Additionally, our observations suggest that a transient in one decayed branch can trigger (or assist with triggering of) a transient in another branch.

Previous studies have shown that subsequent leaders in positive cloud‐to‐ground lightning (+CG) flashes rarely traverse pre‐existing channels to ground. In this paper, we present evidence that this actually can be common, at least for some thunderstorms. Observations of +CG flashes in a supercell storm in Argentina by Córdoba Argentina Marx Meter Array (CAMMA) are presented, in which 54 (64%) of 84 multiple‐stroke +CG flashes had subsequent leaders following a pre‐existing channel to ground. These subsequent positive leaders are found to behave similarly to their negative counterparts, including propagation speeds along pre‐existing channels with a median of 8 × 10⁶ m/s, which is comparable to that of negative dart leaders. Two representative multiple‐stroke +CG flashes are presented and discussed in detail. The observations reported herein call for an update to the traditional explanation of the disparity between positive and negative lightning.