On the evening of September 2nd, 2023 an incredible spectacle of nature occurred in the four corners region of the United States. Rare upward lightning initiated from the iconic natural landmark Shiprock.  Storm photographer Jim Tang (@wxmann on X, @stormchasertang on Instagram) positioned himself 6.8 km southeast of Shiprock in anticipation of lightning passing over this feature, as a cluster of storms with numerous cloud and ground flashes developed to the southwest.  In a span of 13 minutes, Jim witnessed a remarkable five upward lightning flashes initiating from the peak of Shiprock.  He has graciously provided a copy of his images and a description of what he observed to allow me to conduct an analysis of this amazing event.  To capture the images, Jim used a Nikon Z7ii with Nikkor 24-70mm f/4 lens set at 46mm.  ISO was set at 100 and exposure times were 20 seconds.  The aperture ranged between f/8 and f/9.

What follows is a look at the correlated lightning location data recorded during these upward lightning flashes as well as a description of the captured images.  My goal was to analyze the combined data and provide a characterization of the likely physical processes that caused the upward lightning flashes.

I would like to thank Chris Vagasky for providing the lightning location data used in this analysis.

Upward Lightning Flash #1
2023-09-03, 02:37:23 UT

The first upward lightning flash occurred at 2023-09-03, 02:37:23 UT.  The upward lightning channel was non-branched and propagated southwest upon reaching cloud base.  The attachment point to Shiprock was on the south side.  A video was posted on Tiktok that appears to show this lightning flash and the upward propagation of the leader.

The National Lightning Detection Network data recorded nine impulsive events shown below.  The data format is date and time, longitude, latitude, estimated peak current in kiloamps, and type (polarity -/+ followed by IC for cloud flash or CG for ground flash).  The data that is in bold/italic is located within 1 km of Shiprock.

2023-09-03T02:37:22.780Z,-108.8593,36.6084,-8.0 kA, -IC

2023-09-03T02:37:22.780Z,-108.855,36.6179,-3.6 kA,-IC

2023-09-03T02:37:23.379Z,-108.8357,36.6866,-13.8 kA, -CG

2023-09-03T02:37:23.486Z,-108.8352,36.6872,-9.9 kA, -CG

2023-09-03T02:37:23.519Z,-108.8358,36.6866,-7.4 kA, -IC

2023-09-03T02:37:23.552Z,-108.8353,36.6865,-13.7 kA, -CG

2023-09-03T02:37:23.600Z,-109.1555,36.5224,5.3 kA, +IC

2023-09-03T02:37:23.782Z,-109.1102,36.5514,6.5 kA, +IC

2023-09-03T02:37:24.057Z,-109.1097,36.5567,7.2 kA, +IC

Four of these NLDN events had a location within 1 km of Shiprock.  Given that these close events were all negative indicates negative charge being transferred to ground, and therefore an upward positive leader initiated from Shiprock.  These four impulsive events are similar to the dart leader / return stroke sequence sometimes seen with normal downward negative cloud-to-ground lightning.  After the upward positive leader channel decays, a bipolar recoil leader can form with the negative end traveling back down toward ground as a dart leader.  Upon reaching the ground, or in this case the top of Shiprock, a bright return stroke occurs which travels back up the channel.  The return stroke is the impulsive event recorded by the NLDN.  For this first flash, this sequence occurred at least four times with estimated peak currents ranging from 7.4 – 13.8 kA.

Figures 1 and 2 below show all the impulsive events recorded by the NLDN.  Light red circles indicate positive cloud events (+IC) which are incloud impulsive events that lower positive charge, light blue triangles indicate negative cloud events (-IC) incloud impulsive events that lower negative charge, and dark blue triangles indicate negative ground events (-CG) impulsive return strokes which lower negative charge to ground. 

Shiprock is located in the middle of the figure and marked with a brown star.  The yellow circle has a radius of 1 km from Shiprock.  The dark gray and light gray circles have a radius of 10 km and 20 km from Shiprock respectively.  The four negative events close to Shiprock all lie within 200 m of the feature.  The remaining NLDN indicated events are located to the southwest of Shiprock.

Figure 1. NLDN recorded events showing their location relative to Shiprock which is the brown star in the middle.  The yellow circle has a radius of 1km from Shiprock. The dark and light gray circles have radii from Shiprock of 10 km and 20 km respectively.  The light red circles indicate +IC events, the light blue triangles indicate -IC events and the dark blue triangles indicate -CG events.
Figure 2.  Multiple negative NLDN events within 1 km of Shiprock.

Upward Lightning Flash #2
2023-09-03, 02:42:01 UT

The second upward lightning flash occurred at 02:42:01 UT.  This upward positive leader branched once shortly after initiation and was attached to Shiprock north of the previous flash location.  Once again one of the upward positive leaders traveled horizontally southwest upon reaching cloud base.

The NLDN data showed the following

2023-09-03T02:42:01.422Z,-108.8742,36.6966,-4.4 kA,-IC

2023-09-03T02:42:01.423Z,-109.0754,36.5954,14.7 kA, +IC

2023-09-03T02:42:01.757Z,-108.8356,36.6871,-14.6 kA, -CG

For this flash on a single negative subsequent impulsive event was recorded near Shiprock and had an estimated peak current of 14.6 kA.  The two recorded events, which occurred earlier in time were west and southwest of Shiprock.

Figure 3.  NLDN recorded events relative to Shiprock for Flash #2.
Figure 4. NLDN recorded events within 1 km of Shiprock.

Upward Lightning Flash #3
2023-09-03, 02:43:42 UT

The third upward lightning flash, which occurred at 02:43:42 UT, was likely the most spectacular of the five as it exhibited many branches and numerous subsequent impulsive events based on the NLDN data.  The upward leader initiation point was in the same location as the previous flash and the upward leader began branching about mid-way between Shiprock and the cloud base.  Four distinct main branches are visible in the image with numerous branches occurring after the horizontal propagation that took place once the upward leader reached cloud base.  An apparent video of this upward flash was posted on Tiktok, and it shows preceding incloud leader activity (incloud brightening) approaching from the left before the upward leader initiates from Shiprock. The video also shows numerous rebrightening events on the various leader branches characteristic of recoil leaders and dart leader / return sequences.

The NLDN recorded a total of eight events, all within 5 km of Shiprock. Six of these were within 1 km.  Estimated peak current ranged between 3.7 kA and 23.9 kA.

2023-09-03T02:43:42.682Z,-108.8316,36.672,-3.7 kA, -IC

2023-09-03T02:43:42.729Z,-108.8381,36.6865,-6.3 kA, -CG

2023-09-03T02:43:43.024Z,-108.8365,36.6874,-23.9 kA, -CG

2023-09-03T02:43:43.028Z,-108.8357,36.6405,-4.9 kA, -IC

2023-09-03T02:43:43.072Z,-108.8354,36.6875,-8.8 kA, -CG

2023-09-03T02:43:43.107Z,-108.8359,36.6868,-11.0 kA, -CG

2023-09-03T02:43:43.127Z,-108.8361,36.6871,-12.6 kA, -CG

2023-09-03T02:43:43.245Z,-108.8357,36.6874,-9.2 kA, -IC

Figure 5.  NLDN recorded events relative to Shiprock for Flash #3.  All events were negative.
Figure 6.  NLDN recorded events within 1 km of Shiprock for Flash #3.

Upward Lightning Flash #4
2023-09-03, 02:46:40 UT

The fourth upward flash, at 02:46:40 UT, had the same initiation point as the first flash (south of the previous two).  The upward positive leader channel branched once close to cloud base and then branched profusely upon reaching cloud base and transitioning to horizontal propagation.

The NLDN recorded five impulsive events none of which were close to Shiprock.  Interestingly, all the recorded events were positive intracloud flashes located south through southwest of Shiprock between 18 and 30 km.  The visual appearance of the upward leader suggests that it was positive. (Saba et al., 2016)

2023-09-03T02:46:40.667Z,-109.1665,36.5738,21.5 kA, +IC

2023-09-03T02:46:40.672Z,-109.1632,36.5965,4.1kA, +IC

2023-09-03T02:46:41.195Z,-108.8226,36.5277,4.0 kA, +IC

2023-09-03T02:46:41.581Z,-108.7711,36.4997,6.9 kA, +IC

2023-09-03T02:46:41.647Z,-108.7636,36.4958,5.3 kA, +IC

Figure 7.  NLDN recorded events relative to Shiprock for Flash #4.  All recorded events were +ICs.

Upward Lightning Flash #5
2023-09-03, 02:49:55 UT

The fifth and last upward lightning flash, at 02:49:55 UT, had an upward positive leader that appeared to initiate from the same point as flashes 2 and 3 or slightly south of it.  The upward leader’s angled propagation made it difficult to determine.  This upward leader did not branch before reaching cloud base.  Numerous branches formed after the transition to horizontal propagation.

The NLDN recorded three events all between 10-20 km southwest from Shiprock with no impulsive events within 1 km.

2023-09-03T02:49:55.027Z,-108.9875,36.6511,-29.3, -CG

2023-09-03T02:49:55.040Z,-108.9869,36.6508,-7.7, -CG

2023-09-03T02:49:55.302Z,-109.0053,36.6403,-6.3, -IC

Figure 8.  NLDN recorded events relative to Shiprock for Flash #5.  All events were negative.

Below is a plot of all the NLDN indicated events for the five upward flashes.  The grouping of negative events near Shiprock clearly show that upward positive leaders initiated since they are all negative.  The spread of both positive and negative events southwest of Shiprock indicate the direction and location of the charge region favorable for leader propagation.  The mix of positive and negative IC events is typical of impulsive horizontal leader propagation along cloud base.

Figure 9.  All NLDN recorded events for the five upward lightning flashes.

Below is a composite image of Flashes 1, 2, 3 and 5 which show the relative positions of the upward channels for these four flashes.


The foremost objective of this analysis is to determine the likely physical process that led to the five upward lightning flashes.  Our research has shown that lightning-triggered upward lightning is most likely to occur during the summer months and is usually associated with the trailing stratiform precipitation region of mesoscale convective systems.  Specifically, upward leaders from tall objects are triggered by preceding nearby horizontally extensive lightning flashes.  The most common triggering flash type is the positive cloud-to-ground (+CG) flash and negative leader propagation near the tall object either before or after the return stroke is the flash component that creates the electric field change which initiates the upward positive leader most often. (Schumann et al., 2019 and Warner et al., 2012).  However, the +CG return stroke itself can also cause the initiation of an upward positive leader due to the rapid and intense electric field change.  

Self-initiated upward lightning occurs without preceding nearby flash activity, but usually is limited to the cold sector of strong winter storms in the United States (e.g., Warner et al., 2014).  This is due to the lowering of the charge region closer to the tall object and the strong winds which can reduce the shielding corona discharge that may inhibit upward leader initiation.

In looking at the NLDN data for these upward flashes, there is not a clear pattern suggesting +CG flashes were triggering the upward leaders.  In fact, the data seems to show impulsive events primarily due to the development of the upward positive leaders.  The grouping of these NLDN events to the southwest, correlated with the visual confirmation that the upward leaders traveled southwest upon reaching cloud base, suggests that most, if not all, of these more distant events were associated with the upward leaders.  Especially those recorded after a negative event was observed close to Shiprock. It is very common for positive lightning-triggered upward lightning leaders to transition to horizontal propagation upon reaching cloud base (Saba et al., 2016). It is believed that a horizontally extensive shallow region of induced negative charge residing at cloud base (i.e., screening layer of charge) or that negatively charged rain exiting the cloud provides a favorable region for positive leaders to propagate into and through thus resulting in horizontal propagation. Frequently, the development of impulsive recoil leaders occurs along decayed positive leader channel paths after these horizontal leader segments branch profusely.

High-speed video recording showing non-branched upward positive leader transitioning to horizontal propagation and increased branching.

For the last two upward flashes, where there were no negative subsequent events close to Shiprock, the recorded NLDN events to the southwest could either be associated with triggering flash leader activity or activity from the upward leaders.

Jim’s description of the storm evolution was extremely helpful in identifying the likely triggering process.  He observed the storms initially form over the Chuska mountains south-southwest of Shiprock.  After struggling to survive coming off the mountains, they eventually started growing upscale with their anvils extending over Shiprock rock.  He noticed lightning flash activity in the anvil region which prompted him to point a camera in the direction of Shiprock in case something happened even though the primary flash activity was still south-southwest.  Shortly after, upward lightning flashes lit up Shiprock.  A video posted on Tiktok appears to show the storm and lightning activity to the south-southwest of Shiprock prior to the upward flashes.

The description provided by Jim, the Tiktok videos as well as the analysis of the NLDN data suggests that the upward flashes were triggered by nearby lightning activity, however, the triggering flashes were likely cloud flashes instead of more typical +CG flashes.  It is well understood and documented that horizontally extensive +CG flashes in stratiform precipitation regions serve as the dominate triggering process for lightning-triggered upward lightning (Lyons et al., 2014).  However, they are not the only process.  During our research we have observed a few cases in which upward lightning was triggered by negative leaders traveling in the forward anvil region residing over tall objects.  I believe that this was the triggering process for the upward flashes at Shiprock on the evening of Sep 2nd, 2023.  Given the lack of preceding +CG return strokes suggests that intracloud flash activity extending into the anvil positive charge region resulted in incloud negative leaders propagating near Shiprock and causing an electric field change significant enough to initiate upward positive leaders from the feature.  Furthermore, Jim added that, “the stronger-than-normal dynamics associated with this monsoon transition event, and upscale growth/consolidate of multiple updrafts upstream, would also argue for more anvil electricity than the normal desert SW popup storm.” I agree. Below is an animation of how an intracloud flash can trigger upward positive leaders from tall objects.

Intracloud triggering lightning flash results in incloud negative leaders (blue) propagating over tall objects and the initiation of upward positive leaders (red).

Shiprock rises just over 480 m above the relatively flat surrounding terrain.  Like any tall object due to its shape and height, Shiprock enhanced the ambient electric field created by the overlying incloud charge.  The electric field change caused by the approaching incloud negative leaders created favorable conditions for the initiation and development of upward positive leaders from the top of Shiprock.

There are very few examples of upward lightning from natural objects such as prominent rocks and mountains.  Usually, upward lightning occurs from anthropogenic objects such as tall towers and buildings and wind turbines.  Additionally, tall mountains, due to their favorable height and location frequently have communication towers which serve as the initiation point for upward lightning.  Therefore, the observation and analysis of this unique event is an exciting opportunity.  I am very grateful for Jim’s willingness to share his observations and description, and I hope more opportunities present themselves in the future.


Lyons, W. A., T.E. Nelson, T.A. Warner, T.J. Lang, K. Cummins, M. Quick, W. Rison, P. Krehbiel, S.A. Cummer, J. Myers, T. Samaras, P. Samaras, C. Young (2014), Meteorological Aspects of Two Modes of Lightning Triggered Upward Lightning (LTUL) Events in Sprite-Producing MCSs , paper presented at the 23nd International Lightning Detection Conference, Mar 18 – 19, Tucson, Arizona.

Saba, M. M. F., C. Schumann, T. A. Warner, M. A. S. Ferro, A. R. de Paiva, J. Helsdon Jr, and R. E. Orville (2016), Upward lightning flashes characteristics from high-speed videos, J. Geophys. Res. Atmos., 121, doi:10.1002/2016JD025137.

Schumann, C., M. M. F. Saba, T. A. Warner, M. A. S. Ferro, J. H. Helsdon Jr., R. Thomas, and R. E. Orville (2019), On the Triggering Mechanisms of Upward Lightning, Nature Sci Rep., 9, 9576 (2019) doi:10.1038/s41598-019-46122-x

Warner, T. A., T. J. Lang, and W. A. Lyons (2014), Synoptic scale outbreak of self-initiated upward lightning (SIUL) from tall structures during the central U.S. blizzard of 1–2 February 2011, 
J. Geophys. Res. Atmos., 119, doi:10.1002/2014JD021691.

Warner, T. A., K. L. Cummins, and R. E. Orville (2012), Upward lightning observations from towers in Rapid City, South Dakota and comparison with National Lightning Detection Network data, 2004–2010, J. Geophys. Res., 117, D19109, doi:10.1029/2012JD018346.