The Hundred Year Hunt for the Sprite

Walter A. Lyons Yucca Ridge Field Station, FMA Research, Inc. Fort Collins, CO
Russell A. Armstrong Mission Research Corporation, Nashua, NH
E.A. Bering, III Department of Physics, University of Houston, Houston, TX
Earle R. Williams Massachusetts Institute of Technology, Parsons Laboratory, Cambridge, MA

Science often advances at a deliberate and cautious pace. Over 100 years passed before persistent reports of luminous events in the stratosphere and mesosphere associated with tropospheric lightning were accepted by the scientific community. Since 1886, dozens of eyewitness accounts, mostly in obscure meteorological publications, have been accompanied by articles describing meteorological esoterica such as turtles encased in hailstones and toads falling during rain showers. The phenomena were variously described as “cloud-to-space lightning” and “rocket lightning.” A typical description might read, “ In its most typical form it consists of flames appearing to shoot up from the top of the cloud or, if the cloud is out of sight, the flames seem to rise from the horizon.” Such reports were largely ignored by the nascent atmospheric electricity community – even when they were posted by a Nobel Prize winning physicist.

As early as 1925, C.T.R. Wilson proposed possible mechanisms to explain such phenomena. In 1956 Wilson commented, “It is quite possible that a discharge between the top of the cloud and the ionosphere is a normal accompaniment of a lightning discharge to earth...a diffuse discharge between the top of the cloud and the upper atmosphere...many years ago I observed what appeared to be discharges of this kind from a thundercloud below the horizon. They were diffuse, fan-shaped flashes…extending up into a clear sky.” Over the next three decades, several compendia of similar subjective reports from credible witnesses worldwide were prepared by Otha H. Vaughan (NASA) and the late Bernard Vonnegut (The University at Albany). The events were widely dispersed geographically from equatorial regions to above 50° latitude, with 75% occurring over land. The eyewitness descriptions shared one common characteristic - they were perceived as highly atypical of "normal” lightning. The reaction of the atmospheric science community could be summarized as indifference at best.

Then, as so often happens in science, serendipity intervened. Hard Evidence The air of mystery began to dissipate at 0414 UTC on 6 July 1989. Scientists from the University of Minnesota, lead by John R. Winckler, were testing a low-light camera system (LLTV) for an upcoming rocket flight when, quite by accident, they captured two fields of video which provided the hard evidence for what are now called sprites (Franz et al., 1990). From this singular observation emanated a decade of fruitful research into the electrodynamics of the middle atmosphere. In the early 1990s, NASA scientists searched tapes from the Space Shuttle’s LLTV camera archives and confirmed at least 17 apparent sprites above storm clouds. The orbital perspective suggested a relationship between sprites and tropospheric lightning. The specific lightning flashes associated with the observed sprites often were among the brightest in the region. The sprites occurred within milliseconds of the brightest cloud illumination, and were apparently triggered by especially energetic discharges within the storm cell which, while larger its neighbors, had otherwise unexceptional flash rates.

By 1993, NASA developed concerns that this newly discovered “cloud-to-space lightning” might be fairly common and thus pose a potential threat to Space Shuttle missions especially during launch or recovery. Based upon the available evidence, the hunt for these elusive events was directed above the stratiform regions of large mesoscale convective systems (MCSs), known to generate relatively few but often very energetic lightning discharges. On the night of 7 July 1993, a borrowed Xybion LLTV was deployed for the first time at the Yucca Ridge Field Station (near Fort Collins, Colorado). Exploiting an uninterrupted view of the skies above the High Plains, the LLTV was trained above a large nocturnal MCS in Kansas, some 400 km distant. Once again, good fortune intervened as 248 sprites were imaged over the next four hours. Analyses revealed that almost all the sprites were associated with +CG flashes. Within 48 hours, in a totally independent research effort, sprites were imaged from a NASA DC8 aircraft over Iowa. The following summer, the University of Alaska’s flights provided the first color videos detailing the red sprite body with bluish, downward extending tendrils (Sentman et al., 1995). The same series of flights documented the truly bizarre blue jets (Wescott et al., 1998).

During the last seven years the scientific community’s misperception of the middle atmosphere above thunderstorms as “uninteresting” has completely changed. Today a host of phenomena have been named: sprites, blue jets, elves, sprite halos and trolls, with perhaps others remaining to be discovered. Collectively they have been termed Transient Luminous Events (TLEs). But as research has progressed, even this appellation appears somewhat inadequate. The Phenomenology of Transient Luminous Events Much has been learned about the morphology of TLEs.

Sprites can extend vertically between 95 km and under 30 km. While telescopic investigations reveal that individual tendril elements may be of the order of 10 m across, the envelope of the illuminated volume can exceed 104 km3. Sprites almost invariably follow +CG flashes, with time lags of less than one to over 100 ms. To date, there are only two documented cases of sprites associated with negative polarity CGs. The sprite parent +CG peak currents range widely, from under 10 kA to over 100 kA, though on average the sprite +CG peak current is 50% higher than other +CGs in the same storm. High speed video images suggest that sprites usually initiate around 70 to 75 km with both downward and upward development at speeds ~107 m/sec. Sprite luminosity on typical LLTV videos can endure for tens of milliseconds. Photometry suggests, however, that the brightest elements usually persist for a few milliseconds (Armstrong et al., 1998), though occasionally small, bright “hot spots” linger for tens of milliseconds.

By 1995, sprite spectral measurements confirmed the presence of the N2 first positive emission lines. In 1996, photometry provided clear evidence of ionization in some sprites associated with blue emissions within the tendrils and sometimes the sprite body. Peak brightness within sprites is on the order of 1000 kR. In seven years of observations at Yucca Ridge, sprites were typically associated with larger storms (>104 km2 radar echo), especially those exhibiting substantial regions of stratiform precipitation. The TLE-generating phase of High Plains storms averages about three hours. The probability of optical detection of TLEs from the ground in Colorado is highest between 0400 and 0700 UTC. The TLE counts observed from single storm systems has ranged from one to 776, with 48 as an average count. Sustained rates as high as once every 12 seconds have been noted, but more typical intervals are on the order of 2 to 5 minutes. In the early 1990s, Stanford University researchers proposed that the electromagnetic pulse (EMP) from CG flashes could induce a transient glow in the ionosphere between 80 and 100 km altitudes (Taranenko et al., 1993).

Evidence for this was first noted in 1994 using LLTVs at Yucca Ridge, and was confirmed the following year by photometric arrays deployed by Tohoku University (Fukunishi et al., 1996). Elves, as they are now called, are believed to be expanding quasi-torroidal structures which attain an integrated width of several hundred kilometers. Some evidence suggests their intrinsic color is red due to strong N2 first positive emissions. While relatively bright (1000 kR), their duration is <500 microseconds. Roughly 5% of the TLEs detected with standard LLTV imagers are elves. These usually follow by ~300 microseconds very high peak current (often >100 kA) CGs, most of which are positive in polarity. Stanford University researchers, using sensitive photometric arrays, documented the outward and downward expansion of the elve’s disk. They also suggest many more dim elves occur than are detected with conventional LLTVs. These fainter elves have been suggested to be more evenly distributed between positive and negative polarity CGs.

Recently it has been determined that some sprites are preceded by a diffuse disk-shaped glow which lasts about a millisecond and superficially resemble elves. However, these structures, now called “sprite halos,” are less than 100 km wide, and propagate downward from about 85 to 70 km altitude. Columnar sprite elements sometimes emerge from the lower portion of the sprite halo’s concave disk.

Blue jets are the least frequently observed TLE from ground based observatories, in part due to atmospheric scattering of the shorter wavelengths. Evidence from aircraft missions shows blue jets emerging from the tops of electrically active thunderstorms. The jets propagate upwards at speeds of ~100 km/sec reaching terminal altitudes around 40 km. Their estimated brightness is on the order of 1000 kR. Blue jets do not appear associated with specific CG flashes. Curiously, however, CG lightning activity appears to cease for several seconds within a 15 km radius after each blue jet occurrence (Wescott et al., 1998). There have been anecdotal associations of blue jets with hail producing storms. An ER2 pilot over the Dominican Republic flying above Hurricane Georges in 1998 described seeing luminous structures which resembled blue jets.

The troll is the most recent addition to the TLE family. In LLTV videos, trolls superficially resemble blue jets, yet they are clearly dominated by red emissions. Moreover, they occur after an especially vigorous sprite in which tendrils have extended downward to near cloud tops. The trolls exhibit a luminous head leading a faint trail moving upwards initially around 150 km/sec, then gradually decelerating and disappearing by 50 km. It is still not known whether the preceding sprite tendrils actually extend to the physical cloud tops or if the trolls emerge from the storm cloud per se. Figure 1 portrays several forms of TLEs and the lightning within a typical parent MCS. Footnote: (The May-June, 1998 issue of the Journal of Atmospheric and Solar-Terrestrial Physics was dedicated to TLEs and provides a valuable source of references). Parent Storms and Lightning Worldwide, a variety of storm types have been associated with TLEs. These include the larger mid-latitude MCSs tornadic squall lines, tropical deep convection, tropical cyclones and winter snow squalls over the Sea of Japan. It appears, however, that the central U.S. may be home to some of the most prolific TLE producers, even though only a minority of High Plains thunderstorms produce TLEs. Some convective regimes, such as supercells, have yet to be observed producing many TLEs and the few are mostly confined to any stratiform precipitation region which may develop during the late mature and decaying stages. Furthermore the vast majority of +CGs, even many with peak currents above 50 kA, produce neither sprites nor elves which are detectable using standard LLTV systems.

While large peak current +CGs populate both MCSs and supercells, only certain +CGs possess characteristics which generate sprites or elves. Monitoring in the Schumann resonance bands has provided a clue for what differentiates the TLE parent CG from “normal” flashes. Real-time visual sprite observations at Yucca Ridge coordinated with ELF transients (Q-bursts) detected at a Rhode Island receiver station clearly demonstrate that Q-bursts are companions to the +CG flashes generating both sprites and elves (Huang et al., 1999). Sprite parent +CGs are associated with exceptionally large charge moments (300 C-km to >2000 C-km). The sprite +CG ELF waveform spectral color is “red,” that is, peaked toward the fundamental Schumann resonance mode at 8 Hz. As Williams (1998) points out, lightning charge transfers of hundreds of Coulombs may be required for consistency with theories for sprite optical intensity and to account for the ELF Q-burst intensity. Lightning causal to elves has a much flatter (“white”) ELF spectra, and though associated with the very largest +CGs (often >150 kA), exhibits much smaller charge moments (< 300 C-km).

Recent studies of High Plains MCSs confirm that their electrical and lightning characteristics are radically different from the textbook “dipole” thunderstorm model, derived largely from studies of rather small convective storms. Vast horizontal laminae of positive charge are found, often near the 0°C layer, and these structures persist for several hours over spatial scales of ~100 km. With positive charge densities of 1-3 nC/m3, even relatively shallow melting layers (order 500 m) covering 104 to 105 km2 can contain thousands of Coulombs. Some 75 years ago, C.T.R. Wilson postulated that large charge transfers and particularly large charge moments appear to be a necessary condition for conventional breakdown which produces middle atmospheric optical emissions. Sprites occur most readily above MCS stratiform precipitation regions with radar echoes larger than ~104 km2. It is not uncommon to observe rapid fire sequences of sprites propagating above storm tops, apparently in synchrony with a large underlying horizontal lightning discharge. One such “dancer” included a succession of eight individual sprites within 700 ms along a 200 km long corridor. This suggests a propagation speed of the underlying “forcing function” of ~3x105 m/s. This is consistent with the propagation speed of “spider” lightning - vast horizontal dendritic channels tapping extensive charge pools once a +CG channel with a long continuing current becomes established.

It is suspected that only the larger MCS, which contain large stratiform precipitation regions, give rise to the +CGs associated with the spider lightning networks able to lower the necessary charge to ground. The majority of sprite parent +CGs are concentrated in the trailing MCS stratiform regions (Lyons, 1996). The radar reflectivities associated with the parent +CGs are relatively modest, 30-40 dBZ or less. On 18 August 1999, a massive MCC produced a spectacular series of sprites over Nebraska. We identified which CGs were associated with TLEs using data from the National Lightning Detection Network (NLDN). Plotting the CGs by polarity on a GOES satellite infrared cloud top temperature map revealed a distinct pattern (Figure 2). Predominantly –CGs were present along the leading edge of the MCS whereas the +CGs were mostly confined to the highest cloud tops which are usually found above trailing portion of the stratiform precipitation area. Moreover, only a small subregion of the westernmost trailing stratiform area produced sprite and elves. It would appear that this portion of the MCS possessed, for several hours, the requisite dynamical and microphysical processes favorable for the unique electrical discharges which drive TLEs.

The Atmosphere Between the Clouds and Sprites
One of the more important gaps in our knowledge concerns the electrical environment above large thunderstorms during TLE episodes. Consequently, the 1999 Sprite Balloon Campaign conducted three high altitude balloon flights. Flight 3 flew out of Ottumwa, IA between 0039 UTC and 1112 UTC on 21 August 1999. The balloon floated at 32 km and drifted westward at ~30 knots. The payload was instrumented with dual, three-axis electric field detectors, three-axis fluxgate and induction magnetometers, an X-ray scintillation counter, a Geiger-Mueller tube, upward looking high-speed photometer, vertical current density meter, conductivity measurements, and an ambient thermometer. Ground-based LLTV observations were made from three sites (in Wyoming, South Dakota, and Colorado). All three stations had clear skies. There were two small TLE- producing storms, one in eastern South Dakota and one in central Kansas. Of 67 TLEs seen by at least one station or the balloon, five were seen by two or more stations. The balloon data at a typical range of 300 km show that the sprite is accompanied by a positive vertical electric field pulse of ~0.2 V/m. Curiously no perturbation in any component of the electromagnetic fields was observed during the several milliseconds between the lightning flash and a sprite. Also, two very bright elves were detected by ground and balloon optical sensors. The NLDN, however, failed to compute an associated CG. Analysis of the raw network sensor data did, however, reveal the elve parent lightning events were each observed by over 100 sensors. This powerful sferic was so complex as to prevent classification by the NLDN algorithms. Preliminary results from this flight have indicated that more data are required before we understand the complex physics involved. The Physics of Transient Luminous Events TLEs have captured the interest of many theoreticians (Rowland, 1998). Several basic mechanisms have been postulated to explain the observed luminous structures. These include excitation by a quasi-electrostatic (QE) mechanism, sprite production by runaway electrons, and elves from electromagnetic pulses (EMP). More than one process may be operating, but on different temporal and spatial scales, in order to produce the bewildering variety of TLE shapes and sizes.

Absent from almost all theoretical modeling efforts are specific data on key parameters characterizing lightning flashes which actually produce TLEs. Many modelers refer to standard reference texts which, in turn, tend to compile data taken in storm types and locales which are not representative of the nocturnal High Plains. Specifically, many invoke the conventional view that the positive charge reservoir for the lightning is found in the upper portion of the cloud at altitudes of ~10 km. The positive dipole (or tripole) storm model has been found wanting in many midcontinental storms (Williams, 1998). A survey was made of the range of lightning parameters used in over a dozen theoretical modeling studies. While the height of the vertical +CG channel ranges from 4 to 20 km, there is a clear preference for 10 km and above. The amount of charge lowered varies over three orders of magnitude, as does the time scale over which the charge transfer occurs.

Only a few papers consider the possible role of horizontal components of the parent discharge. The charge moment (in C-km), not the peak current as measured by the NLDN, is the key parameter in the basic QE mechanism first proposed by Wilson. The key physics of the problem appear to involve the altitude and magnitude of the removed charge and the time scale on which this occurs – parameters about which little agreement exists. Many theorists note that even with an assumed tall +CG channel (~10 km) this still requires extremely large (~100 Coulombs) charge transfers, typically ten times larger than in “conventional” lightning. Some models yield a thousand-fold enhancement in optical intensity at 75 km for a doubling of lightning charge removal altitude from 5 to 10 km. The use of shorter channels to ground, say 5 km, would imply truly large charge transfers. Yet evidence is accumulating that indeed such may be the case. While the various models simulate optical emissions bearing some (though in many cases rather minimal) resemblance to the observations, such wide ranges in the lightning source term parameters do not appear physically realistic. If in fact such a range of lightning characteristics could produce sprites, why does only a very small subset of +CGs (<1:20 even in active storms) actually produce observable TLEs (with current sensors)? It appears that most models have made assumptions about the lightning in order to produce something resembling a TLE – rather than starting with hard physical constraints on the source term. The reason, of course, is that there is very little data on the actual CGs which generate specific TLE occurrences.

During the 2000 High Plains convection season, many of these questions may begin to be resolved. Taking the Next STEPS. No lightning flash known to produce a sprite or elve has ever been well characterized. To simulate complex TLEs, modelers will require information on the total charge removed and its waveform, the continuing current characteristics, the rate and altitude from which charge was removed, and the geometry of the vertical and especially horizontal lightning channels. Between 22 May and 16 July 2000, a major field observation effort called the Severe Thunderstorm Electrification and Precipitation Study (STEPS) will be undertaken. ( contains a comprehensive overview). The STEPS domain, located 100 to 400 km east-southeast of Yucca Ridge, is ideally situated for acquiring a wide array of optical measurements of TLEs above the storm concurrent with the lightning discharges below. With the inclusion of the 3-D lightning mapping array (LMA) from New Mexico Tech (see,

STEPS will facilitate a wide variety of measurements on TLE characteristics but most importantly that of the unique lightning discharges which give rise to them. STEPS goals will include documenting the differences between +CGs which do and do not produce TLEs. STEPS may confirm whether or not massive “spider” discharges are a necessary condition for sprites, and identify those convective systems capable of producing TLEs. What is the relevance of these findings? It has been suggested that there may be significant production of NOx in the middle atmosphere by sprites. This becomes even more interesting in light of recent observations that regional smoke palls from biomass burns radically enhance the percentage of +CGs within storms, and thus increase sprite counts (and middle atmosphere NOx production?). Schumann resonance analysis methods (Huang et al., 1999) promise a means to obtain a worldwide TLE census. There is growing interest in determining the sources of unusual infrasound emissions detected above sprite-capable MCSs (Al Bedard, NOAA ETL, personal communication). Along with characterizing the optical signatures of TLEs, these findings may have important implications for global monitoring efforts for the Comprehensive Test Ban Treaty. TLEs may contribute in ways not yet understood to the maintenance of the global electrical circuit (Bering et al., 1998). To quantify the impacts of TLEs, we require information on the global frequency (now roughly estimated between one and ten per minute) and their geographic distribution. Some of the STEPS findings may engender some broader issues such as aerospace safety above 15 km.

ACKNOWLEDGEMENTS. This work has been supported by a number of agencies including NASA’s Office of Space Sciences and the Kennedy Space Center, the U.S. Department of Energy, and the National Science Foundation. REFERENCES Armstrong, R.A. J.A. Shorter, M.J. Taylor, D.M. Suszcynsky, W.A. Lyons, and L.S. Jeong, 1998: Photometric measurements in the SPRITES'95 and '96 Campaigns of nitrogen second positive (399.8 nm) and first negative (427.8 nm) emissions. J. Atmos. and Solar-Terrest. Phys. , 60, 787-799. Bering, E.A., III, A. A. Few and J.R. Benbrook, 1998: The Global Electrical Circuit. Physics Today, October, 24-30. Franz, R.C., R.J. Nemzek and J.R. Winckler, 1990: Television image of a large upward electrical discharge above a thunderstorm system. Science, 249, 48-51. Fukunishi, H., Y. Takahashi, M. Kubota, K. Sakanoi, U.S. Inan and W.A. Lyons, 1996: Elves: Lightning-induced transient luminous events in the lower ionosphere. Geophys. Res. Lett., 23, 215-2160. Huang, E., E. Williams, R. Boldi, S. Heckman, W. Lyons, M. Taylor, T. Nelson and C. Wong, 1999: Criteria for sprites and elves based on Schumann resonance observations. J. Geophys. Res., 104, 16943-16964. Lyons, W.A., 1996: Sprite observations above the U.S. High Plains in relation to their parent thunderstorm systems, J. Geophys. Res. 101, 29,641-29,652 Rowland, H.L., 1998: Theories and simulations of elves, sprites and blue jets. J. Atmos. And Solar-Terrestrial Phys., 60, 831-844. Sentman, D.D., E.M. Wescott, D.L. Osborne, D.L. Hampton and M.J. Heavner, 1995: Preliminary results from the Sprites 94 aircraft campaign: 1. Red sprites. Geophys. Res. Lett., 22, 1205-1208. Taranenko, Y.N., U.S. Inan, and T.F. Bell, 1993: The interaction with the lower ionosphere of electromagnetic pulses from lightning: excitation of optical emissions. Geophys. Res. Lett., 20, 2675-2678. Wescott, E.M., D.D. Sentman, M.J. Heavner, D.L. Hampton and O.H. Vaughan, Jr., 1998: Blue Jets: Their relationship to lightning and very large hailfall, and their physical mechanisms for the production. J. Atmos. and Solar-Terrestrial Phys., 60, 713-724. Williams, E.R., 1998: The positive charge reservoir for sprite-producing lightning. J. Atmos. Solar-Terrest. Phys., 60, 689-692