In a previous post I discussed the observation that positive leaders are often visible below cloud base in association with so called “Spider Lightning.” I suspect that these positive leaders form when horizontally extensive negative leader development that propagates just above cloud base decays or becomes cutoff from their initial bipolar development or from the ground termination point in the case of the extensive horizontal negative leader development that frequently follows a +CG return stroke. These positive leaders, which are below cloud base, tend to lag behind or trail the negative leader propagation which is just above cloud base. On some occasions a positive leader associated with this secondary positive leader development will connect with ground resulting in a +CG return stroke. Since the positive leader initiated on part of the original negative leader network that formed, the subsequent return stroke will traverse this previously formed network and cause further extension of the negative leaders (through new negative breakdown in virgin air) once it reaches the outer extents of the negative leader development that formed prior to the return stroke. This can result in a continuation of horizontally extensive negative leader development that travels 10s of kilometers. If another cycle of negative leader cutoff followed by positive leader formation and subsequent +CG return stroke occurs, this horizontal extension of negative leaders can continue for very large distances exceeding 100 km. The resulting field change (or charge moment change) associated with these horizontally extensive flashes can initiate transient luminous events (TLEs) and/or upward positive leaders from tall towers.
On 8/30/11 UT, I was able to record this apparent process with a high-speed camera operating at 10,000 ips. The flash originated to the northeast of my location and a +54.7 kA estimated peak current, +CG return stroke occurred 28 km away at 04:29:11.708 UT based on the NLDN. A standard-speed video camera recorded this correlated return stroke and a sharp brightness increase just outside of the high-speed camera’s field of view also correlated with the return stroke. Horizontal negative leader development following the return stroke propagated just above cloud base towards my location (about 1 km south of Tower 6, see UPLIGHTS post). A few of the leaders were visible just below clouds base and these had the appearance and propagation characteristics of negative leaders. Additionally, electric field sensing equipment located about 5 km to my west recorded a negative field change (atmospheric electricity sign convention) that correlated with the approach of negative leaders. As this development passed over the towers (and overhead the camera) short duration attempted upward leaders were visible from multiple towers. Eventually, weak upward leaders from three towers initiated in close succession (within 7 ms). Two of these leaders exhibited weak recoil leader activity suggesting they were positive polarity. A wide field of view standard-speed camera located 5 km further west than the high-speed camera captured more of the visible negative leaders that emerged just below cloud base as they passed over the towers and the high-speed camera. (See the standard-speed video below).
After the upward leaders decayed and the brightness associated with the horizontal negative leader development decreased, positive leaders were seen to develop downward from multiple locations along the path the negative leaders passed previously. All of the weakly luminous positive leaders had branches that exhibited recoil leader activity. One of the positive leaders connected with the ground at 11.938 UT (in the high-speed camera’s field of view) and the NLDN recorded a corresponding +12.8 kA estimated peak current cloud flash, “+IC” even though there was a clear connection with the ground. The return stroke resulted in a reillumination of the western portion of the original negative leader network path that formed prior to the return stroke, and in fact a negative leader was clearly visible following the return stroke in the same area traversed previously by the horizontal negative leader development. One leader appeared to be new negative leader breakdown in virgin air likely forming a new channel near the previously formed channels. In addition, negative leaders were again visible just below cloud base, but further west than before as seen in the standard-speed video.
As observed frequently with +CG flashes, recoil leaders continued to be active on branches of the downward positive leaders even after the return stroke suggesting these branches were cutoff from the main downward propagating positive leader at the time of the return stroke. These branches did not, therefore, participate in the return stroke (i.e., the return stroke did not travel into these branches during its upward travel from the ground connection point).
The second return stroke did not initiate any upward leaders from the other towers nor did it reinitiate upward leaders from those towers that previously developed upward leaders.
Below is the high-speed camera recording from this flash.
Ron Thomas at New Mexico Tech, gave me an LMA animation showing extensive horizontal negative leader development with 4 sequential +CG return strokes that trailed behind the VHF sources (leading edge of the negative leader development). I suspect that this flash was similar to the one presented here in that positive leaders formed on cutoff ends of the negative leader development and connected with the ground forming +CGs in trail of the preceding negative leaders.
Furthermore, Carey  discussed an LDAR II’s depiction of a horizontally extensive flash in which the “long-lived, spatially extensive, and horizontally stratified lightning channels are clearly reminiscent of the spider lightning activity observed by Mazur et al.,  in stratiform precipitation as part of the intracloud component of a positive CG lightning flash.” He described that the LDAR II recorded VHF sources for one segment as becoming noisy and spatially incoherent in the area of the previously identified channel segment (i.e., there were previously coherent VHF sources that first traveled along the segment). This was followed by a +CG return stroke after which the sources become more spatially coherent and spatially extensive. I believe his description illustrates the initial horizontal negative leader development (first coherent sources that form the channel segment), the subsequent recoil leader activity associated with the positive leaders that form on the cutoff negative leaders (noisy and spatially incoherent sources generated by the spatially separated and non-coherent initiation and propagation of the negative polarity end of recoil leaders that form on cutoff positive leader branches), the +CG as one of the positive leaders connects with ground, and the expansion of new negative leader development following the +CG return stroke.
Carey, L. D., M. J. Murphy, T. L. McCormick, and N. W. S. Demetriades (2005), Lightning location relative to storm structure in a leading-line, trailing-stratiform mesoscale convective system, J. Geophys. Res., 110, D03105, doi: 10.1029/2003JD004371.
Mazur, V., X. Shao, and P. R. Krehbiel (1998), ‘‘Spider’’ lightning in intracloud and positive cloud-to-ground, J. Geophys. Res., 103(D16), 19,811 –19,822.