GLM top megaflash frequency in the (a),(b) North American and (c),(d) South American hotspots. The top megaflash FED from Fig. 1 is plotted in (a) and (c), while the number of unique thunder days with exceptional megaflash activity (megaflash days) is plotted in (b) and (d).
Histograms of exceptional megaflash activity in the (a),(c),(e) North American and (b),(d),(f) South American hotspots. The number of (a),(b) top megaflash days and the top megaflash frequency by (c),(d) month and (e),(f) local hour are plotted for all years (light blue) and individually for each year.
Despite these limitations, GLM has recorded exceptional cases of lightning from across the Americas that are beyond the scale of any flash mapped by an LMA. Lang et al. (2017) described two LMA flashes that were previously certified by the World Meteorological Organization (WMO) as global lightning extremes. The LMA flash with the largest extent was 321 km across while the LMA flash with the longest duration lasted for 7.74 s. The GOES-16 GLM routinely observes flashes on the same scale as these LMA records. Lyons et al. (2020) examined one case of a GLM flash that was over 500 km across, while Peterson (2019a) identified flashes that reached 673 km in extent and 13.4 s in duration. The overall top GLM megaflashes reach 709 km in horizontal extent and 16.7 s in duration. These two GLM flashes have recently been certified by the WMO as new lightning extremes (Peterson et al. 2020b). The new GLM records more than double the previous LMA-based records, and the flashes between the two sets of records represent a population of lightning flashes that has not previously been resolved. Given the interest in these large megaflashes for both public safety and physical lightning research, we aim to answer the question of where and when these largest and longest-lasting flashes in the Americas can be found.
A total of 1,770 unique GLM megaflashes are identified in the GOES-16 dataset between 1 January 2018 and 15 January 2020 that exceeded the Lang et al. (2017) LMA records. These exceptional cases represent the top 0.9% of the 194,880 megaflashes in this dataset. These GLM all megaflash and GLM top megaflash datasets have been published at Peterson (2020a). In total, 1,208 GLM flashes exceed the 321-km LMA distance record, 913 GLM flashes exceed the 7.74-s LMA duration record, and 368 GLM flashes exceed both LMA records. While oceanic lightning is more prone to lateral development than land-based lightning (Peterson et al. 2017a), only a few single megaflash cases reach this exceptional scale. We thus limit our analysis to observations from the GOES-16 GLM, whose domain features complete coverage of the two hotspots for megaflash activity: the Great Plains in North America and the La Plata basin in South America.
The following sections detail how the GLM software delineates individual flashes, how postprocessing corrections are applied to prevent artificial megaflash splitting, how flash extent and duration are measured, and how nonlightning artifacts are removed from the sample.
The primary caveat with using flash extent to approximate flash size is that megaflashes do not progress from the first group to the last in a straight line, and the meandering and branching structure of the flash is not taken into account with this approach. Flash length can be estimated from the GLM data by totaling the incremental distances measured between subsequent groups in the flash. The total incremental group separation for the 709-km top extent flash was 11,124 km while the total group separation for the 16.7-s top duration flash was 12,511 km. The total group separations are 16 and 25 times larger than the extents reported for each flash. These numbers overestimate flash length because subsequent reillumination of the lightning channel will add to the flash length estimate. Thus, flashes with longer durations and more groups (like the top duration flash with 9,593 groups) will have longer total group separation distances than flashes with fewer groups (such as the top extent flash with 7,293 groups).
Solar artifacts are a common source of false detections for GLM megaflash cases. Glint off of bodies of water as well as solar intrusion directly into the instrument optics can cause large numbers of false events that become clustered into large-extent or long-duration flashes. To identify extraordinary megaflash cases, a collection of filters was developed to specifically differentiate between extreme lightning and episodes of solar contamination.
The second filter is the frequency-domain filter described in Peterson (2020b). This filter transforms the flash optical energy time series into the frequency domain and then looks for a low-frequency peak in measured energy that is consistent with solar illumination. Natural lightning lacks such a peak, and this filter allows lightning megaflashes to be differentiated from large-scale solar illumination that makes it through the first filter.
These two automated filters reduce the number of possible GLM megaflash cases that exceed the Lang et al. (2017) records to a reasonable number for manual evaluation. The remaining cases are plotted on top of Advanced Baseline Imager (ABI; Schmit et al. 2017) infrared cloud imagery, and flashes that occur over clear-air regions or whose group-level structure is not consistent with the lateral development of natural lightning (i.e., random group positions or paths that are too linear) are discarded. Applying all of these filters yields the 1,770 GLM flashes noted previously that exceed the former LMA records.
All other regions within the GLM Field of View produced few, if any, exceptional megaflashes over the 2-yr period. Exceptional megaflashes over the Gulf of Mexico, off the west coast of Central America, in the Amazon, or in the Atlantic Ocean occur sporadically at a rate of
Histograms of top megaflash extent and duration are shown in Figs. 1b and 1c. These distributions group the megaflash cases into 10-km or 0.1-s bins and then draw one point on the vertical axis for every flash in a given bin. The precise extents and durations are additionally listed above or below their corresponding points in Figs. 1b and 1c for the top 10 flashes in each category. The top megaflashes certified by the WMO included one flash case whose extent reached 709 km between groups, and another flash that lasted for 16.73 s. The GLM flash with the second-largest extent (673 km across) was discussed in Peterson (2019a). The 36-km difference in flash extent between these two contenders is notable because the next five candidates were 667, 660, 659, and 657 km across. The difference in extent between these flashes and their next-smallest contender was smaller than a GLM pixel (nominally 8 km), which would have added uncertainty to the assessment of which flash was truly larger.
The top GLM flashes in terms of duration, meanwhile, are separated by margins that are considerably longer than the 2-ms frame integration time of GLM. The top 5 megaflash cases lasted for 16.730, 15.205, 14.708, 14.398, and 14.194 s. The GLM megaflash case certified by the WMO as the lightning duration extreme lasted for 1.5 s longer than the next contender. The previous top GLM case described in Peterson (2019a) was ranked fourth, overall, after improving the reclustering code to fix flashes that were split across file boundaries.
To better resolve the megaflash hotspots, Fig. 2 maps the FED distribution from Fig. 1a over only the North America hotspot (Fig. 2a) and South America hotspot (Fig. 2c) regions. The right side of Fig. 2 computes the number of top megaflash days accumulated over the 2-yr period in these regions. Megaflash days are defined as simply the number of unique calendar days where an exceptional megaflash was observed over a given point on the map. This definition is adapted from the concept of thunder days. The FED maps in Figs. 2a and 2c contain 638 individual megaflashes over North America from 83 megaflash days (41 in 2018, 40 in 2019, and 2 in 2020) and 1,095 megaflashes over South America from 97 megaflash days (51 in 2018, 44 in 2019, and 2 in 2020), which, together, account for 98% of all exceptional GLM megaflashes in the GOES-16 dataset.
Figures 1 and 2 show where the top megaflashes in the Americas occur, but when they occur is equally important. Figure 3 counts the number of top megaflash days (top row) per month, the total number of top megaflashes (middle row) per month, and the total number of top megaflashes per local hour (grouped in 2-h bins in the bottom row) during each year and in total for the North America (left column) and South America (right column) megaflash hotspots. The megaflash days depicted here correspond to the hotspot region as a whole, and thus peak at higher yearly totals than the per-pixel megaflash days shown in Fig. 2. 2b1af7f3a8