Unlike most lightning that initiates in the thunderstorm cloud as a bidirectional and bipolar leader that travels both upward and downward towards oppositely charged regions, upward lightning is unique in that it initiates from a tall object and the unidirectional leader only travels upward towards opposite polarity storm charge or a preceding triggering lightning flash component. Lightning-triggered upward lightning (LTUL) is caused by a nearby triggering lightning flash which has one of its components (either leader activity or a return stroke) pass close enough to the tall object to cause a large and rapid electric field change which in turn initiates a self-propagating upward leader from the object. Self-initiated upward lightning (SIUL) does not require a preceding nearby triggering lightning flash. Instead, the electric field due to storm charge generation within the cloud reaches a point at which a self-propagating leader initiates spontaneously. However, in this case the storm charge region is usually much lower and closer to the tall object and sometimes even envelopes the object. In both cases, the shape and height of the tower enhances the electric field locally near the tip so that ionization of the air and resulting leader formation takes place much easier than that over flat ground. In essence, if the tall objects (i.e., towers, wind turbines or buildings) where not there, the upward lightning would not occur.
We have researched upward lightning in Rapid City since 2004, and our findings show that the 10 tall towers along the ridge that runs through the city all have experience upward lightning. During the summer, we only observed lightning-triggered upward lightning and during intense winter storms with heavy snow and strong winds, we only observed self-initiated upward lightning. During the summer months from 2004 through 2014 we recorded recorded 122 upward flashes from the towers all of which were LTUL.
However, during the winter months, we only documented upward flashes during two major snow events. The first and most intense was the devastating blizzard of 4 Oct 2013. During a 21 hour period, the towers in Rapid City initiated 25 SIUL flashes. In addition, the South Dakota Public Broadcasting tower near Faith, South Dakota experienced 17 SIUL flashes. Although we focus our research during the summer months, we just happened to have an electric field meter and digital interferometer operating during the blizzard. The challenge with observing SIUL during heavy snow is that you cannot see the towers because they are obscured by the snow and low clouds. So you have to record the lightning by some other means. The electric field meter recorded the ambient electric field 5 km west of the towers, and the digital interferometer, 23 km east of the towers, mapped lightning leader activity in two dimensions (azimuth and elevation). The interferometer recorded five upward flashes before it lost power along with most of western South Dakota. Below is a video animation of the data recorded by the digital interferometer for one of the upward flashes. You can visualize that you are standing east of Rapid City looking west toward the towers. Each of the individual data points represents the azimuth and elevation to electromagnetic radiation generated by the lightning leader (and received by the sensor) as the leader propagated. The system records data in sequential 4 microsecond windows and determines the direction to the strongest signal in each time window. Since lightning tends to branch as it grows, you see the source points plot the spreading branched leaders as they grow. The leader clearly initiates from a single point and then spreads upward as it branches. Occasionally, you can see a rapid succession of source points that travel back along a branch toward the tower. These are recoil leaders which form on decayed branches in an attempt to reionize the branch.
And here is some video taken from my house during one of the upward flashes.
The only other time that we documented self-initiated upward lightning from the towers in Rapid City was during a strong snow event on Christmas Day 2016. There were three confirmed upward flashes.
So if it is snowing really hard in Rapid City and you hear thunder, chances are you can blame the towers.
If you would like to learn more about lightning, please visit my Education section
Below are journal and conference paper citations on the subject.
Asakawa, A., K. Miyake, S. Yokoyama, T. Shindo, T. Yokota, and T. Sakai (1997), Two types of lightning discharges to a high stack on the coast of the Sea of Japan in winter, IEEE Trans. Power Delivery, 12, 1222–1231.
Bech, J., N. Pineda, T. Rigo, and M. Aran (2013), Remote sensing analysis of a Mediterranean thundersnow and low-altitude heavy snowfall event, J. Atmos. Res., 123, 305-322, doi:10.1016/j.atmosres.2012.06.021.
Brook, M., M. Nakano, and P. Krehbiel (1982), The electrical structure of the Hokuriku winter thunderstorm, J. Geophys. Res., 87, 1207– 1215.
Heidler, F., M. Manhardt, and K. Stimper (2014), Self-Initiated and Other-Triggered Positive Upward Lightning Measured at the Peissenberg Tower, Germany, paper presented at the 2014 International Conference on Lightning Protection (ICLP), 13 – 17 Oct, Shanghai, China.
Lyons, W. A., T. E. Nelson, T. A. Warner, A. Ballweber, R. Lueck, T. J. Lang, S. A. Cummer, M. M. F. Saba, C. Schumann, K. L. Cummins, N. Beavis, S. A. Rudtledge, T. A. Samaras, P. Samaras and C. Young (2014), Large Scale Outbreaks of Thundersnow and Self-Initiated Upward Lightning (SIUL) During Two Blizzards, paper presented at the 23nd International Lightning Meteorology Conference, Mar 20 – 21, Tucson, Arizona.
Pineda, N., J. Figueras i Ventura, D. Romero, A. Mostajabi, M. Azadifar, A. Sunjerga, F. Rachidi, M. Rubinstein, J. Montanyà, O. van der Velde, P. Altube, N. Besic, J. Grazioli, U. Germann, and E. R. Williams (2019), Meteorological aspects of self-initiated upward lightning at the Säntis tower (Switzerland), J. Geophys. Res., doi: 10.1029/2019JD030834
Rakov, V. A., and M. A. Uman (2003), Winter lightning in Japan, in Lightning: Physics and Effects, chap. 8, pp. 308– 320, Cambridge Univ. Press, Cambridge, U. K.
Schultz, C. J., Lang, T. J., Bruning, E. C., Calhoun, K. M., Harkema, S., & Curtis, N. (2018). Characteristics of lightning within electrified snowfall events using lightning mapping arrays, J. Geophys. Res. Atmos., 123. doi: 10.1002/2017JD027821
Wang, D., N. Takagi, T. Watanabe, H. Sakurano, and M. Hashimoto (2008), Observed characteristics of upward leaders that are initiated from a windmill and its lightning protection tower, Geophys. Res. Lett., 35, L02803, doi:10.1029/2007GL032136.
Wang D. and N. Takagi, (2010), Characteristics of winter lightning that occurred on a windmill and its lightning protection tower in Japan, Proceedings of the 3rd International Symposium on Winter Lightning, Jun 13-15, Tokyo, Japan.
Wang D., and N. Takagi, Y. Takaki (2010), A comparison between self-triggered and other-triggered upward lightning discharges, Proceedings of the 30th International Conference on Lightning Protection, Sep 13-17, Cagliari, Italy.
Wang D., and N. Takagi (2012), Three Unusual Upward Positive Lightning Triggered by Other Nearby Lightning Discharge Activity, paper presented at the 22nd International Lightning Detection Conference, 2 – 3 April, Broomfield, Colorado, USA
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.
On May 13th, my daughter and I went out to chase storms that were forming over the Black Hills. A nice cluster of storms moved over Sturgis, South Dakota (home of the Sturgis Motorcycle Rally), and we filmed some close flashes as the storms passed over us. We then followed the cluster toward Bear Butte which is an isolated uplifted hill on the east side of the Black Hills, northeast of Sturgis.
Our primary target decayed and so we focused on new storms that had formed over the Black Hills and were moving directly toward us. They put down some nice CGs, and as they reached us, I repositioned to have Bear Butte in my field of view. A few minutes later we were treated to two spectacular CG lightning flashes directly in front of us and close. They were very bright and very loud. I suspected they were +CGs given their long duration continuing current and exceptional brightness. The Black Hills area and Northern High Plains for that matter exhibits an atypically high percentage of +CG flashes, and trying to understand and explain this anomoly was part of a study I was involved in during the UPLIGHTS research campaign.
For the first flash, I had my infrared triggered cameras set to f/8 and ISO100 in aperture priority mode. Although this setting is ideal for the average CG flash between 5-15 km, the LCD image review showed significant saturation. I reset the aperture to f/11 and the second flash was still somewhat saturated.
Below is the image for the first flash. You will notice there is two CG channels, one in front of Bear Butte and one beyond.
National Lightning Detection Data provided by Vaisala, Inc. indicated the closer CG was in fact positive (electrons traveled upward along the channel) with an impressive 159.6 kA estimated peak current. It struck 2.5 km away. NLDN data indicated the second channel was also a positive CG 12.6 km away and had an estimated peak current of 58.4 kA.
The second flash which is shown below only had one CG termination point.
NLDN data indicated it was a +CG, 2.2 km away with a peak current of 143.1 kA.
Positive CG flashes tend to exhibit higher peak current compared to negative CGs on average and usually do not have multiple return strokes. If my memory serves, I believe the latest published scientific literature has the average peak current for -CGs around 30 kA and 50 kA for +CGs. So these flashes were exceptionally strong. Unlike what we were taught in school, they DO NOT always originate from the top of a thunderstorm or anvil area and DO NOT always strike away from the main storm and rain area. It all depends on where the charge regions form, and in the Northern High Plains, we see a lot of storms with inverted charge regions, which leads to more +CGs. In the near future, I will be adding an education section on my blog which explains this in more detail.
Below is video of the two flashes captured on a Panasonic HPX-170 at 1280x720p60 which uses a global shutter (no annoying rolling shutter artifacts). In the slow playback you will see an artifact on the frame preceding the return stroke. This is saturating brightness bleed over from the subsequent return stroke that occurs in the following frame. After the CCD records a frame, the voltage values from each photosite (which corresponds to each pixel in the image) are shifted to an adjacent storage photosite that is covered. The voltage is then read out from the covered storage photosites while the next exposure is taking place in the non-covered photosites. If the non-covered photosites experience a saturating brightness, some of the voltage can bleed over into the adjacent storage photosites during their readout adding a voltage increase to their recorded values. Since the covered photosites are readout row by row with the data shifting up the CCD array to higher covered photosites after each row is read, the artifact will usually show up lower in the image as the “image data” from the previous frame has moved up when the saturating brightness occurs. These artifacts are often misidentified as attempted leaders that occur close to the camera, when in fact they are only “ghost images” of the bright return stroke channel that occurs in the subsequent frame but shows up on the previous frame (forward in time…que Twilight Zone music.)
You will also notice the integrated recoil leader activity associated with descending positive leaders in the distant second CG during the first flash. This integrated recoil leader activity is a clear identifying characteristic of positive leaders, and I explain this in the previous post.
Below are some additional images from flashes we captured before the storm moved over Bear Butte.
On the evening of 16 July 2012, a weak cluster of storms moved north over Rapid City, South Dakota. A single visible rainshaft formed on the leading edge of the approaching development. At the time of the rainshaft formation, there was no lightning activity along the leading edge. However, lightning flashes were visible to the distant south in the more active trailing portion of the storms. At 04:20:35, (17 July 2012) UT two digital still cameras captured a ground flash near the rainshaft. This was the first visible flash along the leading edge. One camera, a Canon 5D2 Mark III, captured the image using a 16 mm lens set at f/2.8 using ISO 800 and an exposure time of 11 sec. This camera was capturing continuous 11 sec exposures for a timelapse sequence. A second camera, a Canon 7D, captured the image using a 20 mm lens set at f/8 using ISO 100 and an exposure time of 30 sec.
The captured images, which show the entire flash due to the long exposure times, showed a unique feature that I have not seen previously with any flash images that I have captured. The visible channels below cloud base show that there was a main vertical channel that connected with ground and a branch that propagated somewhat horizontally to the left and did not connect with ground. This second branch appeared to propagate toward the rainshaft and upon entering the rain, spread out vertically in both directions while branching extensively. The change in propagation direction and increase in branching appears isolated to inside the rainshaft, and is not apparent on any other channel segments.
An analysis of National Lightning Detection Network (NLDN) data revealed the NLDN recorded a corresponding 6.8 kA estimated peak current, negative cloud to ground stroke (-CG) 8 km southwest of the cameras. This location correlated in both time and direction, and all other preceding NLDN-indicated flash activity was south of the area by 20 km.
I believe that this image provides evidence that a negative leader branch propagated into a positively charged rainshaft that served as a positive potential well favorable for negative leader propagation (Coleman et al., 2003 and Coleman et al., 2008).
Coleman, L. M., T. C. Marshall, M. Stolzenburg, T. Hamlin, P. R. Krehbiel, W. Rison, and R. J. Thomas (2003), Effects of charge and electrostatic potential on lightning propagation, J. Geophys. Res., 108(D9), 4298, doi:10.1029/2002JD002718.
Coleman, L. M., M. Stolzenburg, T. C. Marshall, and M. Stanley (2008), Horizontal lightning propagation, preliminary breakdown, and electric potential in New Mexico thunderstorms, J. Geophys. Res., 113, D09208, doi:10.1029/2007JD009459.
On the night of 12 Apr 2012, the Bay Area of California experienced a storm that literally lit up the skies with upward lightning. Some iconic photographs and video were taken during this event which provided evidence that numerous tall objects developed upward leaders in response to nearby flashes. The foremost images that illustrated what happened that night were taken by Phil McGrew. He had his Canon 5D Mark III camera running continuously using 20 sec exposures. During two of these exposures, his camera captured upward leaders that developed from the Bay Bridge and additional structures on the east side of the Bay. For each of the two photographs that he posted, it is likely that all the upward leaders developed during the same flash that probably lasted less than one second. He was located in a tall building on the east side of San Francisco downtown looking east along the Bay Bridge.
Below are embedding links as provided by Phil’s Flickr page where he has posted two images. Click on the images to go to his Flickr page. The exif data on his Flickr page indicates that the first of the two images was taken at 8:38:29 pm PDT using ISO 100, f/10, 20 sec exposure and a 28 mm lens. In this image there appears to be 5 upward leaders from the Bay Bridge structure and 2 upward leaders from two separate structures on the east side of the bay (likely in the Oakland area).
Again based on the exif data, the second image that Phil captured was at 8:42:41 pm PDT (4 min and 12 sec later) and used the same camera settings. This image (which has rightfully received international acclaim) appears to show 6 upward leaders from the Bay Bridge structure and 4 additional leaders beyond the Bay Bridge likely from structures on the east side of the bay.
Phil’s photographs indicate they were separated by 4 min and 12 sec. Not know the time accuracy of Phil’s camera, we compared the indicated times and time difference between the two images with National Lightning Detection Network (NLDN) data. Based on previous research findings, we suspected that these upward leaders were triggered by positive ground flashes (+CG) within 50 km of the Bay Bridge. Two very large estimated peak current +CG strokes were recorded at 3:39:59.425 and 3:44:12.332 pm PDT. They had estimated peak currents of +129.8 kA and +270.7 kA respectively and were separated by 4 min and 12.907 sec. There was a +27.8 kA stroke at 03:39:22.773 (37 min earlier of the first big +CG) and 2 -CG strokes at 03:40:32.843 and 03:42:21.957 fell within the time spaning the two large +CGs.
Below are GIS plots of the NLDN indicated return strokes and cloud events. The first figure shows the event location, event type by symbol (see legend) and estimated peak current based on relative symbol size. Notice the size of the +CG return stroke symbols relative to the other events.
The next figure shows the NLDN event locations and their times.
The last figure shows the NLDN events and a label of the estimated peak current.
We suspect that the upward leaders that developed from the Bay Bridge were positive polarity and developed following the large estimated peak current positive cloud-to-ground return strokes that occurred inside the Bay. These are examples of lightning triggered upward lightning in which the field change resulting from a preceding flash causes the development of upward leaders from nearby tall objects.
There were a number of other images from other people that showed upward leaders from tall objects during this same night and the other locations included the Golden Gate Bridge and tall buildings in Oakland. We suspect that these upward leaders also developed during the same triggering flashes that caused the upward leaders to develop from the Bay Bridge.
Over the past six years my research colleagues and I have filmed lightning using high-speed digital cameras. In total we have captured 776 naturally occurring lightning flashes with recording speeds as high as 100,000 images per second. 158 of these flashes were cloud flashes in which some of the lightning leaders propagated outside of the clouds. 372 of theses flashes were negative cloud-to-ground flashes (-CG) and 206 were positive cloud-to-ground flashes (+CG). 41 of the flashes were upward flashes originating from tall towers in Rapid City.
During this last summer, we pursued a storm into the Badlands of South Dakota. The Badlands are a beautiful area of erosion in the plains creating incredibly photogenic landscapes, and it is personally one of my favorite places to photograph lightning. On this particular day, I was filming from the Pinnacles Overlook looking east across a road. I filmed a number of flashes, but during one instance I not only captured a +CG flash, I also captured a rare wild tourist roaming the South Dakota plains. Because I film from a highly modified truck with cameras and gadgets sticking out of it, he was a bit curious by the appearance of my vehicle. However, he was clearly more interested in getting to the next viewpoint and quickly scurried off never to be seen again. Here is the video…
Below is a video on the creation of Lichtenberg figures. Interesting is the subsequent bright short discharges that continue to take place after the initial discharge. These seem similar in appearance to recoil leaders, which form on positive leaders branches that become cutoff from a main channel. Compare the two videos below.
YouTube video of Lichtenberg creation.
Upward lightning (upward positive leaders) from a tower filmed at 9,000 images per second.
On the night of 8/24/11, a leading-line/trailing stratiform mesoscale convective system developed and moved over Toronto, Canada. The heart of the trailing stratiform region passed directly over the 553 m tall CN Tower and the people of Toronto were treated to an incredible light show as the tower unleashed at least 34 upward flashes over the span of an hour. Wilke and Elizabeth See-Tho graciously provided me some video of the event and my analysis suggests that all of the upward flashes were triggered by preceding flash activity (lightning-triggered lightning) similar to what I observe in Rapid City, South Dakota. For each case there was clearly in-cloud flash activity that preceded the upward leader initiation. In addition, recoil leaders were visible in a large majority of the upward leaders suggesting they were positive polarity.
Below is a composite image where I stacked selected images from the See-Tho’s video. As you can see, the CN Tower was literally ablaze with lightning leaders over the span of the storm.
Below is the edited video provided by the See-Tho’s. This version plays in real time showing all 34 upward flashes and one spider lightning flash.
Below is the the same video sped up.
Below is video of each flash played at normal speed and in slow motion (total runtime 34 min).
Although I have not obtained nor analyzed lightning data for this storm, I suspect that a majority of the upward flashes were triggered by a preceding +CG flash within 50 km of the tower. Horizontally extensive positive charge regions that form in the trailing stratiform regions of MCSs serve as potential wells for negative leaders that can travel upwards of 100 km. This horizontally extensive negative leader development can take place during an intracloud flash and/or following a +CG return stroke. The negative field change (atmospheric electricity sign convention) experienced at a tall tower by the approach of negative leaders or nearby +CG return stroke can initiate upward propagating positive leaders. The conditions apparently were ideal for this triggering process and weather radar shows this was likely the case.
Below is a radar loop (base reflectivity, 0.5 degree tilt) of the storm development and passage over the CN Tower spanning from 00:02 – 03:41 UT, 8/25/11. The See-Tho’s stated that the first upward flash was shortly after 02:00 UT. This places the leading line convective region just east of the CN Tower with the tower in an area of decrease reflectivity between 30-40 dBz. The tower would stay under this level of reflectivity (i.e., the trailing stratiform precipitation area) until 03:41 UT. The last upward flash the See-Tho’s recorded was at approximately 03:06, but they thought there were a few more upward flashes that followed after they stopped filming.
This truly was a perfect storm to produce upward lightning flashes. I suspect that many transient luminous events (TLEs) in the form of halos and/or sprites may have also been produced by the very same triggering flashes responsible for initiating the upward leaders. The CN Tower is instrumented to measure current through the tower and there is an array of optical sensors including a high-speed camera within 3 km of the tower. Hopefully, all the instrumentation was operational and an outstanding data set was captured.