National Lightning Detection Network Research Articles

Subionospheric VLF signatures and their association with sprites observed during EuroSprite-2003

In this study, VLF observations during EuroSprite-2003 are analyzed in connection with many sprites observed above thunderstorms in central France. The sprites were detected with a sensitive camera from the Observatoire du Pic du Midi in the Pyrenees overlooking storms monitored by the French national lightning detection network. The VLF observations were made in Crete, Greece with a narrowband receiver, and in Nançay, France with a broadband receiver. The storms were in the vicinity of a VLF transmitter (HWV) at Le Blanc, France, whose signal was received on Crete, arriving over a great circle path that cut through the storms to the southeast. The Nançay broadband receiver was located near HWV to the northeast of the transmitter. This setup provided a unique observational set for investigation. The receiver in Crete observed early VLF perturbations in nearly one-to-one association with the sprites, which endorses the findings of earlier work based on EuroSprite-2003 observations from a single storm. While part of the sprite-related VLF perturbations were of the early/fast type, many classified as “early/slow” having onset durations up to ∼2 s and thus suggesting a new mechanism at work which may cause a slow build up of ionization after a sprite. The only elve in the data set was found to associate also with an early/fast VLF perturbation. Moreover, the analysis showed basically no early VLF events to occur in relation to the numerous ±CG discharges that did not lead to sprites. Bandpass filtering of the broadband VLF signal revealed that only about 5% of the sprites were escorted by early VLF perturbations, possibly due to backscatter. Finally, by using all 131 sprites captured during EuroSprite-2003, the time lags of the sprites to the preceding +CG discharges were computed and analyzed. The time-lag distribution had a well defined tail suggesting that at least one third of the sprites observed were lagging the +CG discharges by more than 30 up to 300 ms. In addition these “long-delayed” sprites were not accompanied by any radio-sferics during the sprite observation period, in sharp contrast to the short-delayed sprites which were escorted nearly always by enhanced, burst-like, sferic activity. These observations endorse the notion of long delayed sprites reported in past studies, but also show that their occurrence is much more frequent than it was thought before.

An evaluation of the performance characteristics of the U.S. National Lightning Detection Network in Florida using rocket‐triggered lightning

The performance characteristics of the U.S. National Lightning Detection Network (NLDN) were evaluated using rocket‐triggered lightning data, acquired in the summers of 2001–2003 at the International Center for Lightning Research and Testing (ICLRT), at Camp Blanding, Florida. During the 3‐year period, 37 flashes, containing a total of 159 (158 negative and 1 positive) strokes, were triggered at Camp Blanding. Flashes consisting of the initial stage only (having no return strokes) were not considered in this study. Directly measured currents were obtained for 122 of these strokes in 29 flashes. Camp Blanding and NLDN events were correlated using GPS time stamps. The NLDN recorded 95 Camp Blanding strokes in 31 flashes. Of these 95, usable directly measured currents were obtained for 70 strokes in 22 flashes. Flash and stroke detection efficiencies were estimated to be about 84% and 60%, respectively. Median location error was about 600 m, with larger location errors (greater than 2 km) being associated with strokes having smaller peak currents (5–10 kA). The NLDN tended to underestimate peak current, with the median peak current estimation error (signed) being about 18%. There was a steady trend of improved stroke detection efficiency from 2001 to 2003.

Corona-Discharge-Initiated Mine Explosions

Strong circumstantial evidence suggests that lightning has initiated methane explosions in abandoned and sealed areas of underground coal mines. The Mine Safety and Health Administration (MSHA) investigated several of these occurrences within recent years. The investigated explosions occurred at significant depths, ranging from 700 ft to 1200 ft. Data from the national lightning detection network indicated a strong correlation between the times and locations of the explosions with those of specific lightning strikes. This paper proposes that corona discharge from a steel borehole casing is the most likely mechanism responsible for these ignitions. A recently investigated mine explosion and fire at a depth greater than 1000 ft was selected for this study. Computer simulations were performed, using data collected at the mine site. CDEGS/spl trade/ software from Safe Engineering Services & Technologies, Ltd (SES) and MaxwellSV/spl trade/ from Ansoft Corporation were used for the simulations.

Warm Season Lightning Distributions over the Northern Gulf of Mexico Coast and Their Relation to Synoptic-Scale and Mesoscale Environments

Abstract Cloud-to-ground lightning data from the National Lightning Detection Network are used to create a warm season (May–September) lightning climatology for the northern Gulf of Mexico coast for the 14-yr period 1989–2002. Each day is placed into one of five flow regimes based on the orientation of the low-level flow with respect to the coastline. This determination is made using the vector mean 1000–700-hPa wind data at Lake Charles and Slidell, Louisiana. Flash densities are calculated for daily, hourly, and nocturnal periods. Spatial patterns of composite 24-h and nocturnal flash density indicate that lightning decreases in an east-to-west direction over the region. Flash densities for the 24-h period are greatest over land near the coast, with relative maxima located near Houston, Texas; Lake Charles, Baton Rouge, and New Orleans, Louisiana; Biloxi, Mississippi; and Mobile, Alabama. Flash densities during the nocturnal period are greatest over the coastal waters. Lightning across the northern Gulf coast is closely related to the prevailing low-level synoptic flow, which controls the sea breeze, the dominant forcing mechanism during the warm season. Southwest flow, the most unstable and humid of the five regimes, exhibits the most flashes. In this case, sea-breeze-induced convection is located slightly inland from the coast. Northeast flow, the driest and most stable of the regimes, exhibits the least amount of lightning. The large-scale flow restricts the sea breeze to near the coastline. Geographic features and local mesoscale circulations are found to affect lightning across the region. Geographic features include lakes, bays, marshes, swamps, and coastline orientations. Thermal circulations associated with these features interact with the main sea breeze to produce complex lightning patterns over the area.

Cloud-to-Ground Lightning Production in Strongly Forced, Low-Instability Convective Lines Associated with Damaging Wind

AbstractDuring 9–11 November 1998 and 9–10 March 2002, two similar convective lines moved across the central and eastern United States. Both convective lines initiated over the southern plains along strong surface-based cold fronts in moderately unstable environments. Both lines were initially associated with cloud-to-ground (CG) lightning, as detected by the National Lightning Detection Network, and both events met the criteria to be classified as derechos, producing swaths of widespread damaging wind. After moving into areas of marginal, if any, instability over the upper Midwest, CG lightning production ceased or nearly ceased, although the damaging winds continued. The 9 March 2002 line experienced a second phase of frequent CG lightning farther east over the mid-Atlantic states. Analysis of these two events shows that the production of CG lightning was sensitive to the occurrence and vertical distribution of instability. Periods with frequent CG lightning were associated with sufficient instability within the lower mixed-phase region of the cloud (i.e., the temperature range approximately between −10° and −20°C), a lifting condensation level warmer than −10°C, and an equilibrium level colder than −20°C. Periods with little or no CG lightning possessed limited, if any, instability in the lower mixed-phase region. The current Storm Prediction Center guidelines for forecasting these convective lines are presented.

Lightning Induced Disturbances in Buried Cables—Part II: Experiment and Model Validation

This paper presents experimental results obtained at the International Center for Lightning Research and Testing (ICLRT) at Camp Blanding, Florida during the summers of 2002 and 2003. Currents induced by triggered and natural lightning events were measured at the terminations of a buried power cable, in the cable shield, and in the inner cable conductor. Measurements of the horizontal component of the magnetic field above the ground surface for both natural and triggered lightning are also presented. For distant natural lightning events, locations of ground strike points were determined using the U.S. National Lightning Detection Network (NLDN). Based on the theoretical developments presented in Part I of this paper , the field-to-buried cable coupling equations are solved in both the time domain and in the frequency domain. The obtained experimental results are then used to validate the numerical simulations provided by the relevant developed codes.

The North Alabama Lightning Mapping Array: Recent severe storm observations and future prospects

The North Alabama Lightning Mapping Array became operational in November 2001 as a principal component of a severe weather test bed to infuse new science and technology into the short-term forecasting of severe and hazardous weather, principally within nearby National Weather Service forecast offices. Since the installation of the LMA, it has measured the total lightning activity of a large number of severe weather events, including three supercell tornado outbreaks, two supercell hailstorm events, and numerous microburst-producing storms and ordinary non-severe thunderstorms. The key components of evolving storm morphology examined are the time rate-of-change (temporal trending) of storm convective and precipitation characteristics that can be diagnosed in real-time using NEXRAD WSR-88D Doppler radar (echo growth and decay, precipitation structures and velocity features, outflow boundaries), LMA (total lightning flash rate and its trend) and National Lightning Detection Network (cloud-to-ground lightning, its polarity and trends). For example, in a transitional season supercell tornado outbreak, peak total flash rates for typical supercells in Tennessee reached 70–100 min −1 and increases in the total flash rate occurred during storm intensification as much as 20–25 min prior to at least some of the tornadoes. The most intense total flash rate measured during this outbreak (over 800 flashes min −1) occurred in a storm in Alabama. In the case of a severe summertime pulse thunderstorm in North Alabama, the peak total flash rate reached 300 min −1, with a strong increase in total lightning evident some 9 min before damaging winds were observed at the surface. In this paper, we provide a sampling of LMA observations and products during severe weather events to illustrate the capability of the system, and discuss the prospects for improving the short-term forecasting of convective weather using total lightning data.

A Method to Identify the Optimal Areal Unit for NLDN Cloud-to-Ground Lightning Flash Data Analysis

Abstract Currently, no uniform method exists for determining the optimal areal unit to analyze National Lightning Detection Network (NLDN) data. To address this problem, this paper utilizes the capabilities of modern geographic information systems (GIS) software to develop a consistent method for identifying areal analysis units while considering the location accuracy of the NLDN. Five grid cells were created at spatial resolutions of 0.1°, 0.2°, 0.3°, 0.4°, and 0.5°. To create cloud-to-ground (CG) lightning strikes, random points were generated at nine densities ranging from 1 to 9 strikes per square kilometer. A buffer of 500 m was placed around each random point to account for the margin of error in NLDN location accuracy. Random points that, when buffered to 500 m, still remained completely within the study region were evaluated as a percentage of all of the strikes to determine accuracy. The greatest accuracy of 95.88% was observed in the 0.5° grid cell at a density of 9 strikes per square kilometer. The lowest accuracy of 80.59% occurred in the 0.1° grid cell at a density of 4 strikes per square kilometer. There was little variation between the accuracies in similar grid cells regardless of their density, suggesting that CG flash density will have little effect on accuracy when selecting spatial resolution. Gains in accuracy diminished as spatial resolution increased. Gains in accuracy between the 0.1° and the 0.2° grid cells are approximately 9%. Gains in accuracy between the 0.4° and the 0.5° grid cells are less than 1%. To achieve 95% accuracy, a spatial resolution of no less than 0.4° is required.

Where Is the Real Cloud-to-Ground Lightning Maximum in North America?

Abstract A local maximum in cloud-to-ground (CG) lightning flash density over northwestern Mexico as measured by the U.S. National Lightning Detection Network is examined in detail. Corrections are derived for the relative detection efficiency of the network in this region that is outside the perimeter of the sensors. The need to adjust the parameters of the signal normalization model used for the network is also documented. New propagation model parameters are employed to derive relative detection efficiency corrections for northwestern Mexico and these are then used to show that the actual CG flash density over the region significantly exceeds that observed in Florida, the area with the highest flash density within the perimeter of the network.

Implications of lightning charge moment changes for sprite initiation

We report impulse lightning charge moment changes (defined as occurring in the first 2 ms after return stroke onset) in all cloud‐to‐ground lightning strokes detected by the National Lightning Detection Network in three storms during which above‐thunderstorm sprite video was recorded. After analyzing strokes that both did and did not produce sprites and carefully accounting for lightning‐sprite delay times, we found that sprite initiation on all three nights is consistent with a sharp charge moment threshold; essentially, all charge moment changes above and none below this threshold produced sprites with short delays (<5 ms) from the source lightning. On two nights this threshold was approximately 600 C km and on the other it was approximately 350 C km. This internight variability is probably due to expected variability in the nighttime mesospheric conductivity, and the thresholds themselves are consistent with predictions of conventional breakdown theory. Additionally, we found only one negative polarity lightning stroke from all three storms that exceeded this threshold, indicating that the rarity of documented sprites produced by negative strokes may be largely explained by the lack of sufficiently big negative strokes in the U.S. High Plains.

Climate, Lightning, and Wildfire in the National Forests of the Southeastern United States: 1989-1998

We compiled the record of wildfires from each national forest unit in the southeastern United States for the period 1989 to 1998, then regressed annual fire numbers against lightning flash densities (as recorded by the National Lightning Detection Network) and measures of climatic drought (seasonal precipitation and Palmer Drought Severity Index [PDSI] values). Lightning-caused fires were most common in Florida, where seasonally-dry winters and high lightning-flash density elevate fire risk. Overall regression models consistently identified fire-season PDSI as the most influential explanatory variable in accounting for the spatial and temporal pattern of fires, whereas the contribution of lightning-flash density to regional patterns of wildfires is muted. Lightning density only enters the overall model when a single outlier forest unit from Florida is included in the regression analysis. With respect to physiographically defined regional regression models, lightning density was only a significant influence in the Coastal Plain-East, which included Florida. Some measure of climatic drought was uniformly present in all other regions, except in the driest portion of the study region, the Ozark-Ouachita Uplands. Elsewhere, geographic variation in effective moisture is highlighted. Drought for periods of up to 12 months is necessary to promote burning in the wetter forests of the Appalachian Highlands, but periods of antecedent drought necessary to trigger widespread wildfires are shorter in the warmer and drier Piedmont and Coastal Plain.

High percentage of positive lightning along the USA west coast

[1] Analyses of the cloud-to-ground (CG) lightning characteristics recorded by the US National Lightning Detection Network (NLDN) along the west coast reveal an average annual percentage of positive CG flashes around 40%, while the average value for the USA is approximately 10%. The primary goal of the study was to document and suggest reasons for this positive CG anomaly. Through seasonal and monthly storm analysis, it was determined that the high annual percent positive along the coast is the result of the low variability in total CG flashes throughout the year coupled with a high number of positive CG flashes during the winter season which can be directly attributed to the climate of the Pacific Coast. The topography of the region was determined to affect the areal distribution of the percent positive anomaly by restricting the inward extent of the coastal climate. The secondary goal of the study was to determine whether the following meteorological variables were related to the dominant CG polarity in a storm: 1) the variation of charge region heights using the −10°C level as a proxy and 2) the tilting of the charge regions by strong windshear. Statistical analysis showed that the height of the −10°C temperature level is related to the dominant CG polarity in a storm, while the windshear did not show a significant relationship. In addition, analyses showed that the thunderstorms that produced few CG flashes (<6 flashes) contributed most to the total number of positive CG flashes.

Localization of individual lightning discharges via directional and temporal triangulation of sferic measurements at two distant sites

Lightning discharges generate electromagnetic pulses, known as sferics, that propagate through the waveguide formed by the Earth and the ionosphere. The energy in the VLF band can propagate over great distances (∼10,000 km) with low loss (3 dB/Mm) and can therefore be detected far from the source lightning location. Using sferic measurements from VLF receivers at Palmer Station, Antarctica, and Vieques Island, Puerto Rico, the locations of lightning discharges from regions such as South America, North America, and Africa are triangulated. Both sferic arrival azimuth at the two receivers and the time of arrival difference between the receivers are used to find the best estimate of the lightning discharge location. Owing to the extremely wide separation of the two receivers, ambiguities sometimes result in pairing sferics detected at one site to their matching sferics detected at the other site. In these cases a best‐best estimate scheme is used. Simulation results indicate that for the two receivers considered here the most accurate location performance is in South America with an expected median location error of ∼75 km. The system has the worst performance due north of the Vieques Island site, with an expected median location error of ∼2000 km. Triangulation results for South America are shown for the period 7–10 September 2001. Results for 8 September 2001 are compared to cloud‐to‐ground lightning flashes detected by the National Lightning Detection Network. For this day, 49% of flashes were detected with a median error of about 200 km. Results are also compared to lightning discharges detected by the Lightning Imaging Sensor (LIS) aboard the Tropical Rainfall Measuring Mission (TRMM) satellite. For a particular TRMM pass over South America, 20% of flashes detected by LIS were correctly located using the described technique, with a median error of 187 km. The event rate for a localized South American storm is also monitored and documented for the same 4‐day period.

Doppler Radar and Lightning Network Observations of a Severe Outbreak of Tropical Cyclone Tornadoes

Data from a single WSR-88D Doppler radar and the National Lightning Detection Network are used to examine in detail the characteristics of the convective storms that produced a severe tornado outbreak within Tropical Storm Beryl's remnants on 16 August 1994. Comparison of the radar data with reports of tornadoes suggests that only 13 cells produced the 29 tornadoes that were documented in Georgia and the Carolinas on that date. Six of these cells spawned multiple tornadoes, and the radar data confirm the presence of miniature supercells. One of the cells was identifiable on radar for 11 hours, spawning tornadoes over a time period spanning approximately 6.5 hours. Several other tornadic cells also exhibited great longevity, with cell lifetimes greater than ever previously documented in a landfalling tropical cyclone tornado event, and comparable to those found in major midlatitude tornadic supercell outbreaks. Time-height analyses of the three strongest tornadic supercells are presented in order to document storm kinematic structure and to show how these storms appear at different ranges from a WSR-88D radar. In addition, cloud-to-ground (CG) lightning data are examined for the outbreak, the most intense tropical cyclone tornado event studied thus far. Although the tornadic cells were responsible for most of Beryl's CG lightning, flash rates were only weak to moderate, even in the most intense supercells, and in all the tornadic storms the lightning flashes were almost entirely negative in polarity. A few of the single-tornado storms produced no detectable CG lightning at all. In the stronger cells, there is some evidence that CG lightning rates decreased during tornadogenesis, as has been documented before in some midlatitude tornadic storms. A number of the storms spawned tornadoes just after producing their final CG lightning flashes. Surprisingly, both peak currents and positive flash percentages were larger in Beryl s nontornadic storms than in the tornadic ones. Despite some intriguing patterns, the CG lightning behavior in this outbreak remains mostly inconsistent and ambiguous, and offers only secondary value for warning guidance. The present findings argue in favor of the implementation of observing systems capable of continuous monitoring of total lightning activity in storms.

Spectroscopic measurements of NO2 in a Colorado thunderstorm: Determination of the mean production by cloud‐to‐ground lightning flashes

Ground‐based visible absorption spectroscopy of zenith‐scattered sky light was used to measure changes in the column abundance of NO2 during the passage of a thunderstorm over Boulder, Colorado, on September 12, 2002. The measurements showed a 10‐fold rise in slant column NO2 within the thunderstorm cell, consistent with an increase of about 7 ppbv in the lower core updrafts and main precipitation shaft. Mean mixing ratios of about 3 and 1.5 ppbv were inferred for the upper cell and anvil, respectively. These observations, together with measurements from the Denver NEXRAD and the Colorado State University (CSU) CHILL radars, and cloud‐to‐ground (CG) lightning flash data from the National Lightning Detection Network (NLDN), are used to estimate the flash mean production of NOx by CG lightning. The measurements suggest a mean production rate of (5.8 ± 2.9) × 1026 NOx molecules per CG flash, intermediate between the values derived from previous spectroscopic studies and in good agreement with the widely used value of 6.7 × 1026 NOx molecules per CG flash estimated by Price et al. [1997]. This agreement may be fortuitous, however, since the lightning in the present study was about a factor of two less energetic than the mean assumed by Price et al. and the production of (2.3 ± 1.4) × 1017 NOx/J inferred here is significantly larger than the value of 1 × 1017 NOx/J assumed in their calculations. These differences can be reduced, however, if the estimates of Price et al. are recalculated using newer peak current data from the NLDN, and more recently published flash energy production rates of Wang et al. [1998].

A Small Updraft Producing a Fatal Lightning Flash

Just prior to 1900 UTC 25 July 2000, an 18-year-old male was fatally wounded by a lightning flash on the summit of Pikes Peak, Colorado. This case is believed to be unique in that radar and satellite data indicated that the cell that produced the flash was quite shallow and exhibited marginal reflectivity characteristics typically associated with electrified storms. Additionally, the National Lightning Detection Network indicated that this was the first and only cloud-to-ground (CG) flash associated with this convective cell. It is believed the height and isolated nature of the Pikes Peak massif played a role in the initiation of this flash.

Lightning charge moment changes in U.S. High Plains thunderstorms

We report measurements of impulsive (2 ms) lightning charge moment changes in more than 1000 cloud‐to‐ground (CG) return strokes detected by the National Lightning Detection Network in three United States High Plains storms during the Severe Thunderstorm Electrification and Precipitation Study (STEPS) field program of 2000. The positive CG strokes (+CGs) in a mesoscale convective system exhibit an unusual charge moment distribution with a small median and long tail. Analysis suggests the presence of two different classes of +CGs in this MCS, one with small charge moment changes (≤50 C km) and the other containing larger charge moment changes (50–1400 C km). The distributions of negative cloud‐to‐ground stroke charge moment changes are roughly log‐normal in shape with means varying from 17.7 to 36.8 C km. When combined with past measurements these means vary by a factor of 4 between storms, and there is probably not a single distribution that represents well all storms.

The results from the 1999 sprites balloon campaign

A balloon campaign was conducted in summer, 1999, to measure the stratospheric electromagnetic fields associated with sprites. This paper will summarize some of the salient results of this work. The balloon payloads were instrumented with electric field detectors, magnetometers, an upward looking photometer, and other instruments. Ground observations for detection of sprites included low light level TV (LLTV) observations from three sites, Jelm Mt, WY, Bear Mt, SD, and Yucca Ridge, CO. The disagreements between models and these data will be discussed. We have used the photometer data to find TLEs by checking the trace at the times of cloud to ground (CG) strokes reported by the US National Lightning Detection Network (NLDN). In total numbers, the number of −CG transient luminous events (TLEs) (presumably all halos) predominates over the number of +CG TLEs. On the positive side, the TLE events show a current moment threshold. Inclusion of the −CG events that are not seen from the ground raises mesospheric power input estimates by a factor of ∼5–7. Including the mesospheric effect of CGs not associated with TLEs increases this estimate by a factor of 2.

Statistics and properties of transient luminous events found in the 1999 Sprites Balloon Campaign

Historically, the process of transient luminous event (TLE) detection has required an alert human observer on a low light level television (LLTV) monitor, either in real time or playback. The 1999 Sprites Balloon Campaign payloads had all-sky upward looking photometers not sensitive to events below the balloons. The photometer data was examined at the time of National Lightning Detection Network (NLDN) lightening strokes to find TLEs that were missed visually. 3602 events were analyzed in 4.1 h of storm time. Threshold current moments of ∼50 kA km for the positive cloud to ground (+CG) TLEs and ∼−5 kA km for negative cloud to ground (−CG) TLEs are found.

Error analysis for a long‐range lightning monitoring network of ground‐based receivers in Europe

An experimental long‐range lightning detection system consisting of a network (ZEUS) of six ground‐based radio receivers has been deployed in Europe and has been operated since June 2001. The receivers detect and measure the electromagnetic signal emitted by a lightning source in the Very Low Frequency band, between 5 and 15 kHz (sferics), which propagates over thousands of kilometers in the Earth‐ionosphere waveguide. In this study, lightning location retrievals from the ZEUS network are compared against more definitive sources in order to investigate issues on long‐range lightning retrieval accuracy and detection efficiency. The validation study is carried out over three regions: the U.S. East Coast/Northwestern Atlantic, the African continent, and within the network (Spain). Data originate from the U.S. National Lightning Detection Network, the Lighting Imaging Sensor aboard the Tropical Rainfall Measurement Mission Satellite, and the Spanish National Lightning Network. We investigate the nature of the errors involved in the lightning location retrieval, and propose a new approach for moderating some of these effects. Location errors are shown to vary between 40 and 400 km (mode at 220 km) for very long ranges (&gt;5000 km). Within the network, the distance error does not exceed 40 km (mode at 20 km). Random simulation results are shown to be in good agreement with validation data. Error reduction of 12% (or 20 km at long‐range) can be achieved by accommodating for the varying propagation velocities in the sferics wave. The long‐range detection efficiency is shown to have a mode at approximately 20% and a relatively small spread with no obvious deviation from day to night.