1. Introduction
Lightning is a near-constant presence in the atmosphere. Each year, the Global Lightning Dataset (GLD360) network detects more than two billion lightning events, consisting of both in-cloud (IC) pulses and cloud-to-ground (CG) strokes (Vaisala Xweather 2023). Lightning most frequently occurs in the tropics and midlatitudes, but intermittent events have occurred with increasing frequency in the Arctic in recent years (Holzworth et al. 2021).
Lightning is one of the most frequently experienced natural hazards globally. Recent estimates (Holle and López 2003; Cardoso et al. 2014) suggest that between 6000 and 24 000 lightning fatalities occur globally each year, but limited published statistics (Holle 2016) make this number uncertain. In the United States, lightning fatalities have steadily decreased since 1940 and now average around 27 per year (Jensenius et al. 2023).
Lightning causes significant damage to land and property. Insurance claims due to lightning in the United States averaged more than USD 900 million annually between 2017 and 2019 (Insurance Information Institute 2021). Each year, thousands of lightning-triggered wildfires burn millions of acres of land in the United States (National Interagency Coordination Center 2022). Similar statistics are difficult to locate on a global scale, but Canada reports from CAD 600 million to CAD 1 billion in damage and disruption costs (Mills et al. 2010).
Because of the threat to lives and property posed by CG lightning, it is important for the lightning safety and protection communities to know when, where, and how frequently CG lightning occurs. The amount of lightning strikes in the United States each year is reported to be between 25 million (National Weather Service 2023) and 40 million (Centers for Disease Control and Prevention 2023). The source of these values is unclear, but the CDC value is 60% higher than the NWS value. This paper seeks to address the discrepancy in the two values and provide an answer to the title question: How much lightning actually strikes the United States?
In addition, there is a lack of specificity as to the term “strikes.” Colloquially, this term is often used for any lightning, often including IC and CG. However, it is desirable to identify the number of locations where lightning actually contacts the ground. The lightning protection community provides a primary motivation for improved understanding of ground flash density NG and ground strike point density NSG. International lightning protection standards use these numbers for calculating the need and type of protection necessary for structures and many other objects that are vulnerable to lightning. Underestimating these parameters makes the objects more vulnerable to lightning damage, including to people who may be inside them. Overestimating these parameters leads to unnecessary and often costly expenses that users may not be willing or able to afford.
The International Electrotechnical Commission (https://www.iec.ch/homepage) maintains and updates the globally recognized standards for electrical, electronic, and related technologies. In the case of lightning, IEC 62305-2, Edition 2 (International Electrotechnical Commission 2010), identifies the average annual number of dangerous events ND due to lightning flashes influencing a structure to be protected. This parameter ND depends on the thunderstorm activity in the region where the structure is located and the structure’s physical characteristics. A key parameter in the calculation of the value of ND is the local lightning ground flash density NG. The value of ground flash density (given in flashes per square kilometer) could be estimated using a variety of data sources ranging from the most basic measure, thunderstorm days (with an applicable multiplier such as 0.1), to data from ground flash location networks available in many parts of the world.
In the development of IEC 62305-2, Edition 3, the IEC Lightning Protection Technical Committee (TC 81) agreed that ground flash density underestimates the actual number of events by assuming all flashes will have only one ground strike location. A more accurate assessment would be produced by using ground strike point density NSG. CG lightning flashes with multiple strokes may strike the ground in multiple locations, sometimes large distances apart (Stall et al. 2009), as shown in Fig. 1. A variety of methods have been suggested in the literature to estimate NSG. CIGRE TB 549 (International Council on Large Electric Systems 2013) and Rakov (2007) recommend multiplying the ground flash density values by a correction factor of 1.5–1.7. Pédeboy (2012) proposed a technique to determine NSG directly from the data collected from lightning location systems. Rousseau et al. (2019) suggest that where the value of NSG is not obtained from a lightning location system meeting the requirements of IEC 62858 (International Electrotechnical Commission 2019), the value of NG should be multiplied by a factor of 2 to provide a sufficient safety factor in the calculation of relevant risk components. In the present study, we use lightning data times, locations, and estimates of location uncertainty from a well-calibrated ground-based lightning detection network to determine the value of NSG across the contiguous United States (CONUS) and its variation across the country. Forked strokes (Saraiva et al. 2014), which may simultaneously contact two points on the ground separated by tens of meters, are not included, as these are not detectable by current lightning detection techniques.
A multiple-stroke CG lightning flash from Wyoming making contact with the ground in at least four locations. Numbers indicate each stroke in the flash. The image was created by stacking stills taken from a video of the lightning flash. The video was provided by Brendon Melander.
Citation: Bulletin of the American Meteorological Society 105, 3; 10.1175/BAMS-D-22-0241.1
Using data from the U.S. National Lightning Detection Network (NLDN) between 2017 and 2022, we analyze the annual frequency and location of CG lightning over the CONUS, and we discuss the implications of the results, especially related to the lightning protection community.
2. Data and methods
NLDN is a combined magnetic direction finding and time-of-arrival lightning location system (LLS) with more than 100 sensors located in the CONUS (Cummins and Murphy 2009; Zhu et al. 2020; Murphy et al. 2021a). NLDN detects the electromagnetic waves produced by IC and CG lightning and provides solutions to the time, location, peak current, and classification (IC or CG) of each event. The base detection of NLDN lightning events is IC pulses and CG strokes. CG strokes (and any associated IC pulses) can be grouped into CG flashes based on the time and distance criteria given in Murphy et al. (2021a).
NLDN performance is continually reviewed and developed. As improvements to sensor technology and central processing algorithms are released, the network undergoes upgrades (Murphy et al. 2021a). The most recent upgrade was to the central processing system in May 2021 (Murphy et al. 2021b), with the primary objective of improving the accuracy of classification of events that have long been known to present difficulties, such as low-current positive events. Recent evaluations of NLDN performance (Zhu et al. 2020, 2016; Murphy et al. 2021b) are summarized in Table 1.
Performance indicators of NLDN from recent evaluations of the network. Detection efficiency is the percentage of lightning that was detected by the network. Location accuracy is the distance between where the lightning occurred and where the network calculated its location. Classification accuracy is the percentage of lightning that is correctly labeled as IC or CG.
Ground strike points (GSPs) are the ground contact points shared by one or more CG strokes. Because negative CG flashes often produce more than one stroke, negative CG flashes can have one or more GSPs. Positive CG flashes, by contrast, almost always produce just a single stroke and thus a single GSP, although exceptions have been observed (Zhu et al. 2021). One of the main means of observing the number of GSPs produced by CG flashes is with high-speed cameras (e.g., Poelman et al. 2021a,b, 2023). However, ground strike point locations can be calculated from LLS data (Pédeboy and Schulz 2014; Lesejane et al. 2022), provided that the CG stroke detection efficiency and location accuracy are adequate. The performance of NLDN meets IEC standard 62858 for LLS: Both CG stroke detection efficiency and location accuracy are sufficient that NLDN can now resolve different GSPs within CG flashes.
NLDN began reporting GSPs in November 2021. The method of identifying GSPs involves the overlap in confidence ellipses for each CG stroke in a CG flash, similar to Matsui et al. (2019) and Gcaba and Hunt (2022). When there is sufficient overlap of confidence ellipses, two or more CG strokes are considered to have the same GSP. If there is not sufficient overlap, the CG strokes are considered to have different GSPs. An example of how the NLDN GSP algorithm clusters CG strokes from multistroke CG flashes into GSPs is shown in Fig. 2. In this study, after all data between 2017 and 2022 were processed with the same location algorithm, the data were then analyzed using the NLDN GSP algorithm.
An eight-stroke CG flash in California. Each stroke is numbered, and confidence ellipses are in color surrounding the stroke locations. In this example, strokes 1 and 2 have sufficient overlap to be part of the same GSP, strokes 3 and 4 make up a second GSP, strokes 5–7 are a third GSP, and stroke 8 is the final GSP.
Citation: Bulletin of the American Meteorological Society 105, 3; 10.1175/BAMS-D-22-0241.1
3. Sensitivity to IC–CG classification
It goes without saying that the determination of how much cloud-to-ground lightning strikes in any particular place ultimately depends on how well the measurement of lightning differentiates between CG and IC lightning. Because NLDN does not have perfect classification, particularly of IC pulses (Table 1), we performed the following analyses to assess the sensitivity of the lightning count values to the two possible types of misclassification: 1) CG strokes, typically of low amplitude, that are misclassified as IC pulses and 2) IC pulses that are misclassified as CG strokes (also typically of low amplitude). Misclassifications of the abundant positive-polarity IC pulses as CG strokes have been known for a long time (e.g., Cummins et al. 1998). Historically, NLDN applied hard threshold values to low-amplitude positive discharges to reduce the risk of contamination of the positive CG population by misclassified ICs (historical information given in Murphy et al. 2021a).
To consider the effect of CG strokes, usually of low amplitude, that might be misclassified as IC pulses, the NLDN GSP algorithm was run with and without allowing IC pulses to contribute to GSPs. In the analysis, we look at common GSPs between the two runs that have at least one event that is classified as a CG stroke. We assume, on average, that if an IC pulse is sufficiently close in distance to be clustered into a GSP with a CG stroke, then the IC is probably misclassified, although that assumption may pick up some legitimate IC pulses (e.g., due to preliminary breakdown) that satisfy the GSP clustering requirements. Hence, a comparison of the average number of strokes per GSP for this subset provides an upper bound on the rate at which CG strokes are misclassified as IC pulses. In negative GSPs, we find a 4.4% increase in the number of strokes per GSP when IC pulses are included, relative to when they are not. The increase in strokes added to positive GSPs by this analysis is only a negligible 0.001%.
In an effort to quantify the reverse case, IC pulses misclassified as CG strokes, we again used the GSP algorithm but included only events that are classified as CG strokes. In this analysis, the peak-to-zero (PTZ) time and peak current Ipk were used as filters because PTZ and peak current are two of the strongest discriminators between IC and CG lightning, as discussed in Murphy et al. (2021a). If we assume that CG events with |Ipk| below 15 kA and PTZ below 15 μs (a very conservative threshold, based on the earliest direction finder-based system described by Krider et al. 1980) are likely to be misclassified CG strokes, we reduce the number of negative CG GSPs by 4.4% and the number of negative flashes by 3.8%; based on more recent classification algorithm updates, a less conservative PTZ threshold of 10 μs reduces the number of negative flashes and negative GSPs by 1.0%. In the case of positives, the more conservative values of 15 kA and 15 μs reduce the number of positive GSPs by 3.2% and the number of positive flashes by 3.1%, while the less restrictive 15 kA and 10 μs reduce positive GSPs and positive flashes by 1.8%. Overall, then, the sensitivity analyses indicate that our CG flashes and GSP counts, of both polarities, are probably within 5% of the actual values.
In the case of negative CGs, where flashes often have multiple GSPs and multiple strokes, we further anticipate that it is more likely that single-stroke GSPs have a higher probability of being misclassified than multiple-stroke GSPs. Thus, we have also examined the distributions of PTZ and the absolute value of the peak current |Ipk| of the only strokes in single-CG negative GSPs versus the first strokes in multiple-CG negative GSPs. The relative frequencies of |Ipk| and PTZ of these events are shown in Fig. 3. As expected, we find that the only strokes in single-CG negative GSPs have a higher proportion of |Ipk| < 15 kA and PTZ values < 10 μs. Figure 3 is consistent with the sensitivity analysis described in the preceding paragraph.
Histogram comparing the relative frequency of single negative CG strike points and the first stroke of multiple negative CG strike points as a function of their |Ipk| and PTZ. The small spike in the PTZ distributions at a value of 30 μs is due to contributions from sensors whose PTZ value saturates at 30.
Citation: Bulletin of the American Meteorological Society 105, 3; 10.1175/BAMS-D-22-0241.1
4. Results
Between 2017 and 2022, NLDN reported an average of 55.5 million CG strokes associated with 23.4 million CG flashes that struck 36.8 million GSPs. Table 2 shows the yearly values; the interannual variability is interesting and a topic for future study, although the present paper is dedicated to the averages. The average number of GSPs per CG flash compares favorably with values recorded through high-speed camera campaigns (e.g., Poelman et al. 2021a,b, 2023).
Yearly values of NLDN CG strokes, CG flashes, and GSPs over the CONUS. The multiplicity (number of CG strokes per CG flash) and number of GSPs per CG flash are also calculated.
Because of climatological and geographic variations, lightning is not uniformly distributed throughout the United States (Fig. 4). The greatest concentration of lightning is located near the Gulf Coast and southern plains, including Florida, southern Louisiana, southern Mississippi, and portions of Texas, Oklahoma, and Kansas. The lowest concentration of lightning is located in the western United States along the coast of the Pacific Ocean.
Maps showing the distribution of (a) CG stroke density, (b) CG flash density, (c) GSP density, and (d) the number of GSPs per flash from 2017 to 2022.
Citation: Bulletin of the American Meteorological Society 105, 3; 10.1175/BAMS-D-22-0241.1
Figure 4 shows how the parameters described in bulk for the entire country have variations across the 48 states that have high-quality detection by the NLDN. We show these results in four panels as follows:
- 1)CG stroke density (Fig. 4a): The density is over 48 CG strokes per square kilometer per year in several 2-km grid squares along the Gulf Coast and South Atlantic coast along with southern plains, including Florida, southern Louisiana, southern Mississippi, and portions of Texas, Oklahoma, and Kansas. There are important variations in the western states that are often topographically driven. The lowest stroke density is located along the coast of the Pacific Ocean. This map is similar to that shown by Fig. 1b in Holle et al. (2016).
- 2)CG flash density (Fig. 4b): The pattern is very similar to the stroke in Fig. 4a, but the densities are much smaller. Since there are several CG strokes per CG flash, on average, this is to be expected. This map is similar to Fig. 1 in Holle (2014) and Fig. 1b in Holle et al. (2016) and other publications since it is quite invariant with time.
- 3)CG ground strike point density (Fig. 4c): While the pattern is similar to strokes in Fig. 4a and flashes in Fig. 4b, note that the map of the values is now larger than the flashes shown.
- 4)Ratio of ground strike points per flash (Fig. 4d): Nearly all of the United States has a ratio between 1.4 and 1.8. While this result is more or less anticipated, there are two major departures across the country. One is a roughly triangular shape of low ratios in the northern plains, and the other is a highly variable pattern in the western states. In the latter region, the flash density is often quite small so that variations from one grid square to the next are statistical noise. It should also be mentioned that a few grid squares have ratios less than one, due to the subsequent ground strike points being attributed to an adjacent grid square in very low flash rate regions; a larger grid size eliminates this possibility.
5. Discussion
The maps in Fig. 4 show stroke and flash density to be consistent with many prior depictions, that is, the largest density of CG flashes is in the southeastern CONUS, and values decrease gradually to the north and west. However, the pattern of ground contact points per negative CG flash has a minimum in the northern High Plains, where there tend to be thunderstorms with more CG flashes that have single stroke, the occurrence of anomalously electrified storms (Bruning et al. 2014), and the relative abundance of frequent negative ICs that probably make classification errors somewhat more prevalent than elsewhere. Similarly, there are fewer ground contact points per negative CG flash in the western states. Both of these exceptions to the typical ratio of 1.4–1.8 ground contact points per flash should be considered when designs of lightning protection are made in these regions.
The idea of multiplying NG by a factor of 2 ground strike points per CG flash in the absence of NSG raises several issues. Based on the results of the present analysis, that value is probably too high, except in a few locations where the CONUS maps in Fig. 4 show such values to be approached. In fact, quite a few regions have a ratio lower than 1.5 ground strike points per CG flash. However, it should also be emphasized that excessive attention should not be paid to exact numbers for specific locations. In the western states, for example, there is a large variation in the ratio among adjacent 2-km grid squares. Such local variability is due to a small sample size for each grid square that may only have a few storms that occurred in the dataset. In that situation, an average over many adjacent grids should be used.
For locations outside the CONUS, a few general assumptions about ground strike points per CG flash can be made. Most likely, the ratio over a very large scale is on the order of 1.7 ground strike points per CG flash, since the United States has a wide variety of thunderstorm regimes and camera studies are indicating similar values. Nevertheless, regional variations can be expected such as the northern plains anomaly in the United States and the highly variable values in the western states with low flash rates. Unless a camera study or other method has determined a local value of the ratio of ground strike points per CG flash, it may be necessary to multiply NG by a factor of 2 to be on the safe side, albeit incurring potentially large costs for lightning protection.
As international lightning risk assessments evolve, efforts are made by the lightning protection industry to increase their accuracy. One of the improvements in Edition 3 of IEC 62305-2 is to address the number of ground strike points versus flashes to determine the number of dangerous events in lightning risk assessments. Bouquegneau et al. (2012) suggested the ground strike point density to be estimated as twice the flash density where strike point density values are not available. CIGRE TB 549 (International Council on Large Electric Systems 2013) suggests a correction factor of 1.5–1.7 where only flash density is available. The optimum solution is to derive strike point density directly from the lightning location network, as proposed by IEC 62858 (International Electrotechnical Commission 2019).
6. Conclusions
Data from a high-precision LLS were used to determine two critical lightning parameters over the CONUS, the number of CG flashes, and how many ground strike points are associated with them. Such data are important for lightning protection, human safety, and other applications. It was found that an average of 23.4 million CG flashes occur per year and an average of 1.57 ground strike points are associated with CG flashes. These values are generally in agreement with previous ground truth studies. The spatial distribution across the CONUS shows that the ground strikes per CG flash varies with location. While this study has established the baseline for these parameters in the CONUS, global values remain to be determined.
Acknowledgments.
The authors thank Brendon Melander for the video used to make Fig. 1 and Dr. Daile Zhang for her assistance in creating Fig. 1. The authors also wish to thank the reviewers for their thorough and thoughtful reviews of the manuscript.
Data availability statement.
NLDN data were provided by Vaisala Inc. Researchers can request access to NLDN data through the Vaisala Research Data Grant Program at https://www.vaisala.com/en/lp/request-vaisala-lightning-data-research-use.
References
Bouquegneau, C., A. Kern, and A. Rousseau, 2012: Flash density applied to lightning protection standards. Proc. GROUND 2012, Bonito, Brazil, Brazilian Society for Electrical Protection, http://seftim.com/wp-content/uploads/2014/11/Ground_2012_FLASH-DENSITY-APPLIED-TO-LIGHTNING-PROTECTION-STANDARDS.pdf.
Bruning, E. C., S. A. Weiss, and K. M. Calhoun, 2014: Continuous variability in thunderstorm primary electrification and an evaluation of inverted-polarity terminology. Atmos. Res., 135–136, 274–284, https://doi.org/10.1016/j.atmosres.2012.10.009.
Cardoso, I., O. Pinto Jr., I. R. C. A. Pinto, and R. Holle, 2014: Lightning casualty demographics in Brazil and their implications for safety rules. Atmos. Res., 135–136, 374–379, https://doi.org/10.1016/j.atmosres.2012.12.006.
Centers for Disease Control and Prevention, 2023: Natural disasters and severe weather. Centers for Disease Control and Prevention, accessed 1 June 2023, https://www.cdc.gov/disasters/lightning/victimdata.html.
Cummins, K. L., and M. J. Murphy, 2009: An overview of lightning locating systems: History, techniques, and data uses, with an in-depth look at the U.S. NLDN. IEEE Trans. Electromagn. Compat., 51, 499–518, https://doi.org/10.1109/TEMC.2009.2023450.
Cummins, K. L., M. J. Murphy, E. A. Bardo, W. L. Hiscox, R. B. Pyle, and A. E. Pifer, 1998: A combined TOA/MDF technology upgrade of the U.S. National Lightning Detection Network. J. Geophys. Res., 103, 9035–9044, https://doi.org/10.1029/98JD00153.
Gcaba, S., and H. Hunt, 2022: Ground strike point density map of South Africa. 2022 36th Int. Conf. on Lightning Protection, Cape Town, South Africa, Institute of Electrical and Electronics Engineers, 659–663, https://doi.org/10.1109/ICLP56858.2022.9942529.
Holle, R. L., 2014: Diurnal variations of NLDN-reported cloud-to-ground lightning in the United States. Mon. Wea. Rev., 142, 1037–1052, https://doi.org/10.1175/MWR-D-13-00121.1.
Holle, R. L., 2016: A summary of recent national-scale lightning fatality studies. Wea. Climate Soc., 8, 35–42, https://doi.org/10.1175/WCAS-D-15-0032.1.
Holle, R. L., and R. E. López, 2003: A comparison of current lightning death rates in the U.S. with other locations and times. Int. Conf. on Lightning and Static Electricity, Blackpool, United Kingdom, Royal Aeronautical Society, 103-34 KMS.
Holle, R. L., K. L. Cummins, and W. A. Brooks, 2016: Seasonal, monthly, and weekly distributions of NLDN and GLD360 cloud-to-ground lightning. Mon. Wea. Rev., 144, 2855–2870, https://doi.org/10.1175/MWR-D-16-0051.1.
Holzworth, R. H., J. B. Brundell, and M. P. McCarthy, 2021: Lightning in the Arctic. Geophys. Res. Lett., 48, e2020GL091366, https://doi.org/10.1029/2020GL091366.
Insurance Information Institute, 2021: 2021 Insurance Fact Book. Insurance Information Institute, 250 pp.
International Council on Large Electric Systems, 2013: Lightning parameters for engineering applications. Working Group C4.407, CIGRE TB 549, 118 pp., https://ecigre.org/publication/549-lightning-parameters-for-engineering–applications.
International Electrotechnical Commission, 2010: IEC 62305-2:2010: Protection against lightning—Part 2: Risk management. International Electrotechnical Commission, 171 pp., https://webstore.iec.ch/publication/6794.
International Electrotechnical Commission, 2019: IEC 62858:2019: Lightning density based on lightning location systems—General principles. International Electrotechnical Commission, 30 pp., https://webstore.iec.ch/publication/62861.
Jensenius, J. S., R. L. Holle, and M. A. Cooper, 2023: Efforts to reduce lightning casualties in the U.S. through education and awareness. 12th Asia-Pacific Int. Conf. on Lightning, Langkawi, Malaysia, IEEE.
Krider, E. P., R. C. Noggle, A. E. Pifer, and D. L. Vance, 1980: Lightning direction-finding systems for forest fire detection. Bull. Amer. Meteor. Soc., 61, 980–986, https://doi.org/10.1175/1520-0477(1980)061<0980:LDFSFF>2.0.CO;2.
Lesejane, W., H. Hunt, C. Schumann, and R. Ajoodha, 2022: A Bayesian approach to determining ground strike points in LLS data. 2022 36th Int. Conf. on Lightning Protection (ILCP), Cape Town, South Africa, Institute of Electrical and Electronics Engineers, 434–439, https://doi.org/10.1109/ICLP56858.2022.9942602.
Matsui, M., K. Michishita, and S. Yokoyama, 2019: Characteristics of negative flashes with multiple ground strike points located by the Japanese Lightning Detection Network. IEEE Trans. Electromagn. Compat., 61, 751–758, https://doi.org/10.1109/TEMC.2019.2913661.
Mills, B., D. Unrau, L. Pentelow, and K. Spring, 2010: Assessment of lightning-related damage and disruption in Canada. Nat. Hazards, 52, 481–499, https://doi.org/10.1007/s11069-009-9391-2.
Murphy, M. J., J. A. Cramer, and R. K. Said, 2021a: Recent history of upgrades to the U.S. National Lightning Detection Network. J. Atmos. Oceanic Technol., 38, 573–585, https://doi.org/10.1175/JTECH-D-19-0215.1.
Murphy, M. J., R. Said, J. Cramer, and W. Schulz, 2021b: May 2021 update to the central processing system of the U.S. National Lightning Detection Network. 2021 Fall Meeting, New Orleans, LA, Amer. Geophys. Union, Abstract AE35A-1916, https://agu.confex.com/agu/fm21/meetingapp.cgi/Paper/908680.
National Interagency Coordination Center, 2022: Wildland fire summary and statistics annual report 2022. National Interagency Coordination Center, 51 pp., https://www.nifc.gov/sites/default/files/NICC/2-Predictive%20Services/Intelligence/Annual%20Reports/2022/annual_report.2.pdf.
National Weather Service, 2023: Lightning safety tips and resources. National Weather Service, accessed 1 June 2023, https://www.weather.gov/safety/lightning.
Pédeboy, S., 2012: Identification of the multiple ground contacts flashes with lightning location systems. Fourth Int. Lightning Meteorology Conf., Broomfield, CA, Vaisala, https://www.meteorage.com/sites/default/files/inline-files/ILDC2012%20-%20Identification%20of%20the%20Multiple%20Ground%20Contacts%20Flashes%20with%20Lightning%20Location.pdf.
Pédeboy, S., and W. Schulz, 2014: Validation of a ground strike point identification algorithm based on ground truth data. 23rd Int. Lightning Detection Conf., Tucson, AZ, Vaisala, https://www.meteorage.ch/sites/default/files/inline-files/ILDC2014%20-%20Validation%20of%20the%20ground%20strike%20point%20algorithm.pdf.
Poelman, D. R., and Coauthors, 2021a: Global ground strike point characteristics in negative downward lightning flashes—Part 1: Observations. Nat. Hazards Earth Syst. Sci., 21, 1909–1919, https://doi.org/10.5194/nhess-21-1909-2021.
Poelman, D. R., W. Schulz, S. Pedeboy, L. Z. S. Campos, M. Matsui, D. Hill, M. Saba, and H. Hunt, 2021b: Global ground strike point characteristics in negative downward lightning flashes—Part 2: Algorithm validation. Nat. Hazards Earth Syst. Sci., 21, 1921–1933, https://doi.org/10.5194/nhess-21-1921-2021.
Poelman, D. R., H. Kohlmann, W. Schulz, S. Pedeboy, and L. Schwalt, 2023: Ground strike point properties derived from observations of the European Lightning Location System EUCLID. 2023 12th Asia-Pacific Int. Conf. on Lightning (APL), Langkawi, Malaysia, Institute of Electrical and Electronics Engineers, 1–5, https://doi.org/10.1109/APL57308.2023.10182055.
Rakov, V. A., 2007: Lightning phenomenology and parameters important for lightning protection. Ninth Int. Symp. on Lightning Protection (IX SIPDA), Foz do Iguaçu, Brazil, University of São Paulo.
Rousseau, A. S., F. Cruz, S. Pedeboy, and S. Schmitt, 2019: Lightning risk: How to improve the calculation? Int. Colloquium on Lightning and Power Systems, Delft, Netherlands, CIGRE, https://www.meteorage.com/lightning-risk-how-improve-calculation.
Saraiva, A. C. V., L. Z. S. Campos, L. Antunes, O. Pinto Jr., and K. L. Cummins, 2014: Analysis of forked strokes characteristics over southeastern Brasil during the summer season of 2013. 23rd Int. Lightning Detection Conf., Tucson, Arizona, Vaisala, https://www.vaisala.com/sites/default/files/documents/Saraiva-Analysis%20of%20Forked%20Strokes%20Characteristics%20over%20Southeastern%20Brasil-2014-ILDC-ILMC.pdf.
Stall, C. A., K. L. Cummins, E. P. Krider, and J. A. Cramer, 2009: Detecting multiple ground contacts in cloud-to-ground lightning flashes. J. Atmos. Oceanic Technol., 26, 2392–2402, https://doi.org/10.1175/2009JTECHA1278.1.
Vaisala Xweather, 2023: Total lightning statistics 2022: The annual lightning report. Accessed 1 June 2023, https://www.xweather.com/annual-lightning-report.
Zhu, Y., V. A. Rakov, M. D. Tran, and A. Nag, 2016: A study of National Lightning Detection Network responses to natural lightning based on ground truth data acquired at LOG with emphasis on cloud discharge activity. J. Geophys. Res. Atmos., 121, 14 651–14 660, https://doi.org/10.1002/2016JD025574.
Zhu, Y., W. Lyu, J. Cramer, V. Rakov, P. Bitzer, and Z. Ding, 2020: Analysis of location errors of the U.S. National Lightning Detection Network using lightning strikes to towers. J. Geophys. Res. Atmos., 125, e2020JD032530, https://doi.org/10.1029/2020JD032530.
Zhu, Y., and Coauthors, 2021: Multiple strokes along the same channel to ground in positive lightning produced by a supercell. Geophys. Res. Lett., 48, e2021GL096714, https://doi.org/10.1029/2021GL096714.
