Confusion in the interpretation of standard-speed video observations of optical flashes above intense cloud-to-ground lightning discharges has persisted for a number of years. New high-speed (3000 frames per second) image-intensified video recordings are used along with theoretical modeling to elucidate the optical signatures of elves and sprites. In particular, a brief diffuse flash sometimes observed to accompany or precede more structured sprites in standard-speed video is shown to be a normal component of sprite electrical breakdown and to be due entirely to the quasi-electrostatic thundercloud field (sprites), rather than the lightning electromagnetic pulse (elves). These 'sprite halos' are expected to be produced by large charge moment changes occurring over relatively short timescales ( 1 ms), in accordance with their altitude extent of  70 to 85 km. The relatively short duration of this upper, diffuse component of sprites makes it difficult to detect and to discriminate from elves and Rayleigh-scattered light using normal-speed video systems. Modeled photometric array signatures of elves and sprites are contrasted and shown to be consistent with observations, Ionization in the diffuse portion of sprites may be a cause of VLF scattering phenomena known as early/fast VLF events. (24 refs.)

A theoretical model of columniform sprites (or c-sprites), a distinctive class of high altitude, temporally brief optical emissions, is presented and compared to observations which extends earlier work (Symbalisty, Roussel-Dupre, and Yukhimuk, 1998) by making a strong connection with meteors. The key features of the model are: (1) an ambient conductivity profile that falls between a measured nighttime and a measured daytime conductivity; (2) an aerosol reduced conductivity in a trail from a meteor that passed through some time during the evening, and (3) a cloud-to-ground (CG) lightning stroke, with sufficient charge transfer, subsequent to and occurring within an hour of the development of the reduced conductivity trail. The model predicts a temporally brief column of light resulting from the conventional breakdown of air in a strong electric field in the observed altitude range. For the case of a positive CG stroke the emissions are extinguished by the passage of a runaway electron beam. The electron beam is initiated by the same positive CG lightning stroke that allows the high altitude conventional breakdown to occur and propagates from the cloud tops to the ionosphere. Based on these modeling results, a negative CG lightning stroke, for the same amount of charge transfer, produces a column of light about twice as bright. The emissions are extinguished, in this case, by the ambient conductivity taking into account the increase due to the conventional breakdown of air. In both cases, for the CG lightning stroke parameters examined, the simulated c-sprite emissions are brief and last less than 17 ms, or one CCD video field. (47 refs.)

Images of sprites have been recorded at 1 ms resolution revealing several new sprite properties. Sprites appear to occur in a highly structured mesosphere, and the authors suggest that the cause of some of this structure is the sprite activity itself. Evidence is seen in events where a subsequent nearby sprite appears to re-activate the volume of a previous sprite, in sprites where tendrils and branches develop away from the normally observed vertical direction, and in beads that lasts much longer that the parent sprite. Sprites can be large with horizontal widths of more than 40 km and can extend from the clouds up to the lower ionosphere thus affecting a large volume of the atmosphere. The total horizontal area of sprites during one storm over Nebraska was a significant fraction of the area covered by the associated thunderstorm raising the possibility of larger scale measurable mesospheric effects. (18 refs.)

In order to investigate the spatial and temporal variations of sprites and elves and their spectral structures, we have carried out photometric observations during the SPRITES' 97 campaign using two multianode array photometers (MAPs). Each MAP has 5 fields of view arrayed in vertical and a time resolution of 52 mu s, which enables us to detect the rapid vertical motion of sprites/elves. Since the emissions of sprites and elves mainly consist of the 1st and 2nd positive bands of Nsub 2, the intensity ratio of these bands gives us information on the energy distribution of electrons which excite Nsub 2 molecules via collision processes. Thus, one of the MAP instrument with an optical sharp cut filter was used to measure only the Nsub 2 1st positive band emissions in the wavelength range of 560-800 nm, while the other MAP without a filter was used to measure both the Nsub 2 1st and 2nd positive band emissions in the range 350-800 nm. Comparing the data from these two MAPs, we estimated the relative ratio of the 1st/2nd positive bands of Nsub 2. During this campaign, we obtained 66 events data of sprites or elves with the MAPs. It is found that over the entire region of the head of column-shaped sprites the relative ratio of the 1st/2nd positive bands of Nsub 2 is small at the initial phase of the luminosity enhancements lasting only about 1 ms. In the second luminosity enhancement occurring 1-2 ms after the initial phase, the relative ratio of the 1st/2nd positive bands decreases in the lower part of the head. These facts imply that electrons which excite Nsub 2 molecules have higher energy at the initial phase over the entire head, while the energy of electrons is high only at lower altitude in the second enhancement. (8 refs.)

On August 14, 1998, three separate daytime sprite events were detected via a unique extremely low frequency (ELF) sprite signature. The onset of the sprite ELF signatures was delayed by 11.0-13.2 ms from positive cloud-to-ground strokes which had attained exceptionally large charge moment (charge times height) changes of 3900-6100 C.km. It is shown that a charge moment change of 6100 C.km may have been sufficient for conventional breakdown at approximately=54 km altitude, assuming an experimentally measured ion conductivity profile of Holzworth et al. (1985). The daytime sprites themselves contained unusually large charge moment changes of approximately=2890 C.km, approximately=1200 C.km, and approximately=910 C.km. (26 refs.)

Sprites, a form of brief luminous discharge in the upper atmosphere above a thunderstorm, were observed and imaged on two video cameras in Australia's Northern Territory. These were the first such ground-based observations made outside the United States. Sprite discharges typically took place between the altitudes of 50 km and 80 km and spanned an average width of 44 km. Many of the sprite events were of long duration, with an average of 145 ms. These spatial and temporal features were similar to those observed from the ground and the air in the United States. During the longer events, some luminous discharge elements were observed to decay as other new elements formed. As the new elements were often laterally displaced from the old, the sprites sometimes appeared to dance across the sky. This phenomenon has been observed in Colorado and named 'dancing sprites'. The lateral progression of sprite elements observed in the Northern Territory was overwhelmingly in one direction and covered distances of up to 90 km. (20 refs.)

Recent time-resolved multi-color photometric data obtained on one class of lightning-related transient upper-atmospheric electromagnetic events called sprites have confirmed an impulsive ionization emission during the sprite initiation. Data have also been obtained on some sprites which do not exhibit observable tendrils and which exhibit ionization emission that, if present, is below the authors' detection limit. This suggests that some sprite events exhibit strong ionization while others do not. These results indicate that conditions causing sprite optical emissions are highly variable. (16 refs.)

A large scale model of sprites based on a phenomenological probabilistic approach to modeling of streamer corona discharges is developed. The model utilizes the experimentally documented macroscopic properties of positive and negative streamer coronas in air and allows a realistic determination of the propagation of multiple breakdown branches in a self-consistent electric field. The model results reproduce the large scale volumetric shapes of sprites, agree with the experimentally documented thundercloud charge moment changes in sprites producing cloud to ground lightning discharges (CGs), and demonstrate fundamental asymmetries between sprites generated by CGs of positive and negative polarity. (23 refs.)

Video and photometric observations of a meteor-triggered 'jet' event in association with the occurrence of a sprite were collected during the SPRITES '98 campaign. The event raises interest in the question of possible meteoric triggering of upper atmospheric transients as originally suggested by Muller [1995]. The event consisted of three stages: (1) the observation of a moderately bright meteor, (2) the development of a sprite in the immediate vicinity of the meteor as the meteor reached no lower than  70 km altitude, and (3) a slower-forming jet of luminosity that appeared during the late stages of the sprite and propagated back up the ionization trail of the meteor. The event is analyzed in terms of its geometry, its relevance to the meteor and the implications to existing theories for sprite formation. (31 refs.)

High altitude air breakdown, manifested as 'red sprites', is reported in close association with negative cloud-to-ground lightning (-CG) on at least two occasions above an unusual storm on August 29, 1998. Data from high speed photometry, low-light-level video, and receivers of lightning electromagnetic signatures in the frequency range 10 Hz to 20 kHz are used to establish the association and indicate that the causative -CG discharges effected unusually large vertical charge moment changes ( Delta Msub Qv) of up to 1550 C.km in 5 ms. The existence of sprites caused by -CG events, rather than the regularly associated +CG events, has immediate implications for sprite models and observations. (18 refs.)

Red sprites consist of multiple ionised columns extending above a thunderstorm from  30 km to  90 km. Electron densities in these columns are very much larger than the ambient background, perhaps fives orders of magnitude at 70 km. These highly ionized structures cause observable perturbations in subionospheric VLF transmissions known as 'VLF sprites'. Three models of initial sprite electron density are considered, and using a realistic ionization relaxation model the time dependence of electron density is derived. A 3D Born propagation code of is used to compute the time profile of a VLF sprite. Two profiles show good agreement with the time signature experimentally observed, in that scattered amplitude and phase decrease logarithmically with time. These simulations provide insight into the nature and structure of sprite columns, and indicate an additional constraint which should be applied to red sprite creation models. (18 refs.)

We analyze synchronized high speed video images and ELF-VLF radio emissions from 11 sprite clusters observed on 6 October 1997. Quantitative analysis shows that vertical lightning charge moment changes of 150-1100 C.km occurred before the optical emissions reached their peak with delays of 2-11 ms from the lightning discharge. This threshold unexpectedly decreases with increasing delay from parent lightning to peak emissions. We also find that sprite charge moment change and minimum sprite altitude are not well correlated with the vertical charge moment change in the parent discharge. These observations do not agree well with present sprite generation models, and we suggest that streamer development and horizontal lightning charge motion can play a significant role in sprite generation. (12 refs.)

High speed video of sprites show that they are typically initiated at an altitude of about 75 km and usually develop simultaneously upwards and downwards from the point of origin with an initial columniform shape. The initial development of sprites appears to be dominated by corona streamers with velocities in excess of 10sup 7 m/s. Many of the observed characteristics are consistent with a conventional breakdown mechanism for both sprite initiation and initial sprite development. (20 refs.)

In the last decade there has been a great deal of interest in the detection and understanding of phenomena occurring above active thunderstorms. The discovery of the optical phenomena now termed 'red sprites' is discussed, along with the properties that have been experimentally determined. Areas of disagreement between experimentalists are pointed out. Other optical and electromagnetic phenomena associated with red sprites are presented. These include blue jets, transionospheric pulse pairs, and gamma ray flashes. Particular attention is given to the work on perturbations on very low-frequency radio wave transmissions ('VLF sprites'), which has provided estimates of the electrical properties of sprites. Research into activity above thunderstorms will continue to lead to a greater understanding of the coupling between thunderstorms in the troposphere to the stratosphere, mesosphere, ionosphere, and beyond. (109 refs.)

Red sprites were discovered by chance in 1989 when a low-light TV system was pointed above an active thunderstorm. Red sprites are associated with columns of ionization in the Earth-ionosphere waveguide, from above the thunderstorm into the D region of the ionosphere. The ionized columns have been detected through 'VLF sprites', perturbations of the phase and/or amplitude of subionospheric VLF transmissions, which can be used to study the electrical properties of red sprites. There is extensive experimental evidence that VLF sprites may involve wide scattering angles and can produce back scattered radiation. Here we present a numerical and theoretical study of the scattering of subionospheric VLF transmissions caused by the plasma columns associated with red sprites. Comparison of the VLF scattering from sprites is made between a non-Born rigorous model which assumes the sprites are infinite columns of constant conductivity, and a three dimensional Born scattering code. Both formulations show excellent agreement with one another. The formulations predict VLF sprites similar to those experimentally observed for all scattering angles. This shows that the conclusions of previous studies into VLF sprites making use of the non-Born formulations of Rodger et al. [1997] are valid. The modeling provides strong evidence that red sprite plasma is highly ionized in comparison with the ambient nighttime ionosphere, being nearly 5 orders of magnitude greater than the ambient at some heights. (32 refs.)

Ground flashes with positive polarity associated with both sprites and elves excite the Earth's Schumann resonances to amplitudes several times greater than the background resonances. Theoretical predictions for dielectric breakdown in the mesosphere are tested using ELF methods to evaluate vertical charge moments of positive ground flashes. Comparisons of the measured time constants for lightning charge transfer with the electrostatic relaxation time at altitudes of nighttime sprite initiation (50-70 km) generally validate the electrostatic assumption in predictions made initially by Wilson (1925). The measured charge moments (QdS=200-2000 C-km) are large in comparison with ordinary negative lightning but are generally insufficient to account for conventional air breakdown at sprite altitudes. The measured charge moments, however, are sufficient to account for electron runaway breakdown, and the long avalanche length in this mechanism also accounts for the exclusive association of sprites with ground flashes of positive polarity. The association of elves with large peak currents (50-200 kA) measured by the National Lightning Detection Network in a band pass beyond the Schumann resonance range is consistent with an electromagnetic pulse mechanism for these events. (84 refs.)

Broadband ELF/VLF measurements of sferics near Ft. Collins, Colorado, demonstrate that ELF sferic energy is a proxy for sprite occurrence which can be used to estimate the number of sprites produced by a thunderstorm. Ultra-long range ( 12000 km) measurements at Palmer Station, Antarctica, confirm the application of this proxy to storms where no video observations of sprites are available. Comparison with high-resolution photometer measurements demonstrate the simultaneity of sprite luminosity and an ELF 'second pulse' believed to be radiated by electrical currents within the sprite body (Cummer et al., 1998). Measurements of the second ELF pulse are used to identify a quantitative relationship between the current in sprites and total sprite luminosity. (15 refs.)

The results of numerical simulations of red sprite discharges, namely the temporal evolutions of optical emissions, are presented and compared with observations. The simulations are done using the recently recalculated runaway avalanche rates. The temporal evolution of these simulations is in good agreement with ground-based photometer and CCD TV camera observations of red sprites. The authors' model naturally explains the 'hairline' of red sprites as a boundary between the region where the intensity of optical emissions associated with runaway breakdown has a maximum and the region where the intensity of optical emissions caused by conventional breakdown and ambient electron heating has a maximum. Other important characteristics of the simulated sprite such as color, shape, altitude, and intensity are in agreement with the observations as well. The authors also present for the first time simulations of red sprites with a daytime conductivity profile. (20 refs.)

Simultaneously recorded discrete excitations of Earth-ionosphere cavity resonances at Silberborn, Germany, and Hollister, California,  9.1 Mm apart, are used to triangulate source locations of lightning flashes in the continental United States with an accuracy of  0.8 Mm, as verified by the National Lightning Detection Network. The identified lightning flashes are mainly associated with positive cloud-to-ground discharges with first return stroke peak currents  20-70 kA. 80% of these particular lightning flashes are associated with sprites, as verified by simultaneous low-light level TV camera observations at Yucca Ridge, Colorado. This high probability of sprite detection is attributed to particularly large cloud-to-ground lightning currents, simultaneously exciting both Earth-ionosphere cavity resonances and sprites. (17 refs.)

Expressions are developed to characterise the scattering of VLF electromagnetic waves from a finite-length ionised column at some height in the Earth-ionosphere waveguide. This formulation has been examined to investigate the perturbations of VLF transmissions caused by scattering from red sprites (very large luminous phenomena observed over thunderstorms). The idea of a correction factor is advanced to move from a previous model using infinitely long columns to one with truncated columns (50-90 km altitude) inside the waveguide. Using a night-time ionosphere model a quantitative comparison is made between experimentally observed and modelled red sprite scattering characteristics. It is found that the conductivity of the sprite columns (`spritelets') must be about 10sup -4 S msup -1 to generate the experimentally observed scattered fields. This suggests that the electron density of red sprites can be five orders of magnitude greater than the night-time ambient electron density at some heights. (43 refs.)

A synthetic spectrum of red sprites due to electron energization by the electric field from lightning is presented. It is computed by using the electron distribution function obtained from a Fokker-Planck code, which includes various inelastic losses. The model also includes the atmospheric attenuation of the optical emissions. The results are compared with observed red sprite spectra. Some implications of the results to models of red sprites are discussed. (26 refs.)

We have obtained blue (350-475 nm) video images and simultaneous high-time resolution narrow-band blue (415-435 nm) photometry records of four sprite events. The brightest blue images show a sustained tendril geometry and a nearly constant intensity of emission over the entire vertical extent of the sprite (from 35-90 km altitude). Photometer light curves display an exponential decay with a 0.3 ms time constant, a FWHM on the order of 0.1-1.0 ms, and are probably dominated by 427.8 nm (Nsub 2sup + 1N) emission. The data support the observations of Armstrong et al. (1998) and support the contention that significant ionization occurs during sprite generation. (27 refs.)

The authors have obtained blue photometric measurements of the Nsub 2 second positive 399.8 nm and the Nsub 2sup + first negative 427.8 nm emission from sprites, elves and lightning, along with supporting video images. The pulse width and intensity results for sprites are consistent with those of Suszcynsky et al. (1998). The red emission from sprites has been independently and unambiguously identified by Hampton et al. (1996) and Mende et al. (1995) as the nitrogen first positive band. The source has been attributed to electron impact excitation from low energy electrons ( approximately=1 eV) in the sprite. The short pulse width of the 427.8 nm and 399.8 nm photometer time traces obtained in this investigation are probably not from the same source that produces the red emission. The results indicate an initial energetic ionizing event sufficient to ionize and excite nitrogen followed by secondary electron processes which give rise to the dominant red emission. The photometer results for elves are consistent with the EMP mechanism suggested by Inan et al. (1996). The photometer traces obtained for lightning indicate emissions consistent with a `continuing current' as the charge redistributes within the thunderstorm cloud. The authors find that the ratio of the intensity of the 399.8 nm Nsub 2 (2P) emission to that of 427.8 nm Nsub 2sup + (1N) emission can be used to discriminate among sprites, elves and lightning. (41 refs.)

Electromagnetic data recorded in conjunction with the Sprites '95 campaign are presented. The primary data set consists of electric and magnetic field waveforms related to visually identified sprites and elves recorded on the night of 24-25 July 1995. The data were collected near State College, PA, from a mesoscale convective system (MCS) located about 2100 km away near Lubbock, TX. The optical events were visually identified from an observation station in Fort Collins, CO. Presented are the waveforms of the sferics, a description of the measurement system, and a discussion of the signature traits of optical event-producing sferics. All of the sferics recorded which were related to visually identified events exhibited primarily unipolar `slow tail' electromagnetic signatures of order one millisecond duration in the direction indicating positive lightning. Similar waveforms of opposite polarity, indicating ordinary negative lightning, were not accompanied by any observed high altitude optical events. (19 refs.)

In an accompanying paper, Rodger et al. (1998) modelled a red sprite as a set of thin columns (`spritelets'). By examining a semi-random distribution of coupled spritelets, they showed that the scattered field from a VLF radio transmitter has a complex amplitude pattern with deep minima, and sometimes produces backscattering with similar amplitudes to forward scatter. In this special issue, Wescott et al. (1998) have presented the positions of the columns in a sprite event determined by 3D triangulation. Using this distribution of columns and with specific reference to the radio physics observations during sprites campaigns, the present authors show that the conclusions arrived at using a semi-random distribution of columns holds for the experimentally determined distribution. (4 refs.)

VLF phase and amplitude perturbations in `early/fast Trimpis' have been observed simultaneously on two or more transmitters at two or more receiver sites and simultaneously with sprites observed optically in Colorado. In all cases the early/fast Trimpis could apparently be resolved into two components: the RORD (rapid onset, rapid decay) component coincident with the sprite and a `classic Trimpi' component having the slower onset and decay of Trimpis generally thought to be caused by whistler-induced electron precipitation from the radiation belts. The phase and amplitude perturbations can decay at very different rates, and even change in sign from positive to negative perturbation (`overshoot') during decay. This implies a spatial separation between the rapid decay plasma responsible for the RORD and the slow decay plasma responsible for the classic Trimpi component of up to a few tens of km. This is much less than the expected separation of sprite plasma and electron precipitation plasma induced by the same cloud-ground lightning. The discovery of very strong early/fast Trimpis in the tropics (Darwin, Australia), where electron precipitation is unlikely, which showed similar effects like overshoot but much more clearly, caused an examination of recent Colorado data of high resolution. The Colorado Trimpis, which are clearly associated with sprites, show the same features as the Darwin Trimpis where optically detected sprites were not available. Both can be explained in terms of sprite plasma decaying from the bottom up without recourse to electron precipitation. (25 refs.)

This paper reports observations of a distinctive form of sprites associated with positive CG flashes carrying currents of 23 or less to about 100 kA in mesoscale thunderstorms. The sprites are characterized by long vertical columns about 10 km long, less than 1 km in diameter, and show virtually no variation in brightness along their length. Three dimensional triangulation of what the authors define as a `columniform' sprite (c-sprite) event on the evening of 19 June 1995 showed that the individual elements had an average terminal altitude of 86.7 km and an average bottom of 76.2 km. Some show faint diffuse `hair' or tendrils extending above and below the column. The sprite columns are nearly vertical, in video imagery. On some evenings, c-sprites are the dominant form of sprite activity above thunderstorms but, on other nights with many sprites, they may not be observed at all. Comparison of c-sprite forms vs National Lightning Detection Network (NLDN) positive cloud-to-ground current, shows a progression from simple thin vertical forms to brighter and more complicated forms. Theoretical explanations which predict the form and vertical structure of the classical sprites do not at present account for these different forms. (12 refs.)

The close association of mesospheric sprites with positive ground flashes has led to the frequent assumption that positive charge is transferred from the top of a thunderstorm with positive-over-negative charge structure, from an altitude of 10 km or higher. Electrical and meteorological observations are reviewed which support a different picture: sprites are produced by laterally extensive mesoscale convective systems (MCS) in which the positive charge reservoir predominates in the 4-6 km ranges of altitude. The behaviour of the surface electric field during the End-of-Storm Oscillation and the behaviour of the vertical electric field above MCS during positive ground flashes both suggest a predominant in-cloud dipole moment with opposite polarity to that of the ordinary thundercloud. Lightning charge transfers of a few hundred Coulombs from the 4-6 km height range may be required for consistency with theories for sprite optical intensity and to account for ELF Q-burst intensity. (47 refs.)

Video images, and photometric and VLF data, were obtained by the University of Minnesota SKYFLASH system of 38 `sprite' events associated with a strong frontal system located in the upper midwest, U.S.A., on 20-21 June 1996. Besides two image-intensified TV cameras, the SKYFLASH system included telescopic photometers sensitive to Rayleigh scattered lightning flashes by viewing the zenith over the station (O'Brien Observatory, University of Minnesota, about 40 km NE of Minneapolis-St Paul) and also several VLF channels with 300 Hz to 10 kHz bandwidth for recording the electromagnetic `sferics'. The sprites covered a wide range of sizes, from small kilometer-size filaments to huge luminous objects 50-60 km in lateral dimension. All the sprites appeared to consist of bundles of filaments, and always followed-within several ms-a `trigger' could-ground discharge which, in 35 of the 38 events, was positive. It is difficult to find physical mechanisms that explain this positive stroke preference. The larger events reached from 80 km almost to cloud tops, but the small events were localized near 60 km altitude, which is the `bright' region of sprite luminosity, a fact also not well explained theoretically. The filamentary structure of sprites also presents challenges to explain. Of about a dozen lightning storms observed with SKYFLASH in the period from 1993 to 1996 in the upper midwest, only two had an appreciable number of sprites. (23 refs.)

The sequence of videotape observations of the upper atmospheric optical flashes called sprites, jets, starters, and ELVES are described in the successive phases of search, discovery, confirmation, and exploration for the years before 1993. Although there were credible eyewitness accounts from ground observers and pilots, these reports did not inspire a systematic search for hard evidence of such phenomena. The science community would instead wait for serendipitous observations to move the leading edge of this science forward. The phenomenon, now known as a sprite, was first accidentally documented on ground based videotape recordings on the night of 6 July, 1989. Video observations from the space shuttle acquired from 1989-1991 provided 17 additional examples to confirm the existence of the sprite phenomenon. Successful video observations from a mountain ridge by Lyons, starting on 7 July, 1993, and night-time aircraft video observations by Sentman and Wescott on 8 July, 1993 established the basic science of the sprite phenomena by acquiring and analyzing data based on hundreds of new events. The 1994 Sprites campaign and the video entitled 'Red Sprites and Blue Jets' popularized the name sprite and provided a vocabulary of terms to describe the visual attributes. Prior to this video, investigators used a variety of vague descriptive words to describe the individual events. Also, during the 1994 campaign, Wescott and coworkers obtained the first quantitative measurements of jets and provided the name `blue jets'. A third phenomenon was discovered in video from the STS-41 mission (October 1990) in the lower ionosphere directly above an active thunderstorm. It consisted of a large horizontal brightening several hundred kilometers across at the altitude of the airglow layer. In 1995, Lyons and associates confirmed the existence of this type of very brief brightening which they named Emissions of Light and Very Low Frequency Perturbations From Electromagnetic Pulse Sources (ELVES). (48 refs.)

Charge and current systems associated with sprites constitute a part of the large scale atmospheric electric circuit, providing a context for physical understanding of recently discovered ELF radiation originating from currents flowing within the body of sprites. It is shown that the impulse of the electric current driven in the conducting body of the sprite by lightning generated transient quasi-electrostatic fields produces significant electromagnetic radiation in the ELF range of frequencies, comparable to that radiated by the causative lightning discharge. (10 refs.)

Analysis of a large number of VLF scattering events associated with sprites results in an average scattering distribution having a strong frontal lobe, and also having strong large-angle scattering. This is consistent with the scattering pattern produced by a theoretical model of the sprite as an array of interacting vertical plasma columns, indicating that some part of the sprite is 'hard' and contains complex structure (most likely the 'stalactites' seen in optical sprites). Attempting to determine the scattering distribution due to sprites using a high threshold for event detection could result in observation of only the frontal lobe and the minima on either side of it, and not the scattering at larger angles. This problem would be worsened by the use of only amplitude, and not phase, monitoring to detect perturbations, and could lead experimenters to the conclusion that sprites are narrow-angle scatterers, with a smoothly varying diffuse structure. (25 refs.)

A theory of the electrical breakdown (EB) above thunderstorms is developed. The streamer type of the EB is proposed for the explanation of observations of fine spatial structures and bursts of blue optical emissions associated with sprites. (21 refs.)

In July-August, 1996, Stanford University carried out broadband ELF/VLF measurements of the magnetic field radiated by positive cloud-to-ground (CG) discharges associated with Red Sprites. The authors report these measurements for 17 sprite associated discharges that occurred during a 15 minute period on August 1, 1996. The current and charge moments for each of the events are deduced, and it is found that, in every case, intense continuing currents of  1 ms duration are responsible for most of the positive charge transfer to ground that precedes the appearance of the sprite. The time delay between the causative positive discharge and the video field in which the sprite first appeared varied from 0 to 15 ms for the larger events to as much as 100 ms for the smaller events. The authors suggest that in the smaller events the removal of significant positive charge during this delay interval is accomplished through a horizontal intracloud discharge. (15 refs.)

Measurements of ELF-radiating currents associated with sprite-producing lightning discharges exhibit a second current peak simultaneous in time with sprite luminosity, suggesting that the observed ELF radiation is produced by intense electrical currents flowing in the body of the sprite. (16 refs.)

VLF sprites are identified by high angle scattering of VLF transmissions by the highly conducting plasma columns which are observed as the luminous structure of 'red sprites'. By using multiple receivers and/or transmitters the location of the sprite plasma can be determined using existing VLF methods. It is shown that the complex electrical structure of sprites leads to uncertainties in the position determination. However, these are normally no more than the lateral width of the sprite structure itself. (12 refs.)

A new and improved model of red sprites is presented. Emphasis is placed in accounting for the puzzling observation of the spatial structure in the red sprite's optical emissions. The model relies upon a horizontal fractal lightning discharge, which generates the EMPs that excites the optical emissions in the lower ionosphere. It is shown that the fractal model may account for the observed sprite's spatially structured optical pattern, while reducing the typical charge threshold to approximately 100 C. (18 refs.)

We use the runaway air breakdown model of upward discharges to calculate optical, radio, and X-ray radiation generated by red sprites. Red sprites are high altitude (up to 90 km) lightning discharges. Aircraft based observations (Sentman et al., 1995) show that sprites are predominantly red in color at altitudes above  55 km with faint blue tendrils, which extend downward to an altitude of 40 km; the duration of a single sprite is less than 17 ms, their maximum brightness is about 600 kR, and estimated total optical energy is about 1-5 kJ per event. The ground based observations show similar results, and provide some additional information on spatial and temporal structure of sprites (Winckler et al., 1996), and on sprite locations (Lyons et al., 1996). One difference between aircraft and ground-based observations is that blue tendrils are rarely observed from the ground. Sprites usually occur above the anvils of large mesoscale convective systems and correlate with strong positive cloud to ground discharge (Boccipicio et al., 1995). Upward discharges are the most probable source of X-ray emission observed above large thunderstorm complexes by the Compton Gamma-ray Observatory (Fishman et al., 1994). To escape the atmosphere these gamma -rays must originate above 25 km altitude. Red sprites are usually observed at altitudes higher than 50 km, and are therefore a likely source of this X-ray emission. (9 refs.)

Recent years of observations of the upper atmosphere and the lower ionosphere brought a fascinating collection of new phenomena including optical, radio, and gamma-ray emissions originating in the 20 to 90 km altitude range. Up to now, the most diverse phenomenology has emerged from the optical observations which have led to the identification of red sprites, blue jets, blue starters and elves. Most of the previous studies have concentrated on relating such phenomena in the upper atmosphere to regular lightning discharges in the troposphere. For example, sprites and jets are believed to be optical manifestations of electrical discharges in the upper atmosphere caused by quasi-electrostatic fields penetrating to high altitudes during a regular lightning discharge. The sprite/jet discharge itself can be caused by the runaway air breakdown or regular air breakdown. The standard theory for optical airglow transients in the lower ionosphere above the thunderstorms also known as elves suggests that they are produced during interaction of electromagnetic pulses (EMP) from lightning with the lower ionosphere. Heating of the ambient electrons by the EMP in the D-region can result in excitation of optical emissions once the optical excitation thresholds are reached. In this paper the authors suggest that in addition to this mechanism elves can be caused by an EMP generated by sprites and jets. If sprites and jets are indeed accompanied by electrical discharges then some energy of their EMPs reaches to the ionosphere and heats ambient electrons there that in turn stimulates optical emissions similar to EMPs from regular lightning. (10 refs.)

Optical and search coil magnetometer data obtained from the SPRITES'96 campaign carried out at Yucca Ridge Field Station, Colorado in July 1996 have presented clear evidence for the excitation of ULF transients with their dominant power at 1-2 Hz by strong lightning discharges producing elves and sprites. The most striking feature is that the ULF transients exhibit different wave forms in the case of sprites without preceding elves and the case of sprites with preceding elves. In the former case damped, quasi-sinusoidal oscillations commence impulsively at the onset of sprites, while in the latter case quasi-sinusoidal wavelets with a duration of  3 s are excited, and elves and sprites occur within each wavelet. It is likely that these ULF transients are due to the nonlinear excitation of the ionospheric Alfven resonator by strong lightning discharge, as proposed by Sukhorukov and Stubbe [1997]. (16 refs.)

VLF sprites are identified by high angle scattering of VLF transmissions by the conducting plasma columns which appear as the luminous columns of 'red sprites'. VLF sprites are 'early/fast Trimpis' and probably vice versa. Recently discovered properties of early/fast Trimpis are the logarithmic decay of the amplitude of the scattered signal and monotonic variation of its phase. These properties are explained in terms of scattering from a vertical column or set of columns extending from 50 km (or lower) altitude to about 80 km. (15 refs.)

Large area multicell thunderstorms lead to the formation of vertically oriented cylindrical structures of gravity waves at mesospheric altitudes closely resembling those observed in optical emissions associated with transient luminous glows called sprites. (31 refs.)

Transient high altitude optical emissions referred to as 'sprites' are believed to occur as a result of the transfer of large amounts of charge ( 100-300 C) from cloud altitudes of 5-10 km to the ground. Using a general subionospheric ELF propagation model, the authors quantitatively interpret magnetic field waveforms of ELF radio atmospherics originating in mid-western U.S. lightning discharges and observed at Stanford ( 1800 km range) to determine the temporal variation of the lightning current and thereby measure the charge transfer during the stroke. For 6 sprite-producing lightning current waveforms observed on July 24, 1996, the authors find that 25 to 325 coulombs of charge was transferred during the first 5 ms of the discharges, assuming a 10 km altitude for the initial charge. (19 refs.)

ELF waveforms at large distances from strong CG (cloud-to-ground) discharges are evaluated in the framework of Greifinger's night-time propagation model. It is shown that if the CG discharges triggering red sprites involve, as now generally accepted, >sub  10sup 2 C of charge then they should generate remote ELF atmospherics with considerably larger magnitudes than measured in the upper ELF (>300 Hz) range. (39 refs.)

A synthetic spectrum of red sprites due to electron energization by the electric field from lightning is computed by using the electron energy spectrum obtained from a Fokker-Planck code, which includes various inelastic losses. The results are compared with observed sprite spectra. Implications to models of red sprites are presented. (20 refs.)

Transient luminous events (sprites, blue jets, elves) above large mesoscale convective systems (MCSs) over the U.S. High Plains have been routinely monitored from the Yucca Ridge Field Station near Fort Collins, Colorado using ground-based low-light video systems. The author analyzed 36 sprites above the Nebraska MCS of August 6, 1994. The results lend further support to the hypothesis that sprites are almost uniquely associated with positive cloud-to-ground lightning flashes (+CGs). Sprite-associated +CGs also averaged substantially larger peak currents than the remaining +CG population (81 kA versus 30 kA in this storm system). There is some evidence that sprite-associated +CGs also have higher stroke multiplicity. This study yields no evidence of sprites associated with negative CG events. In the central United States an additional requirement appears to be that the parent MCS has a contiguous radar reflectivity area exceeding 20-25,000 kmsup 2. The majority of the sprites occur above the large stratiform precipitation region and not the high-reflectivity convective core of the MCS. Triangulation of a limited number of paired images (from September 7, 1994) suggests that the sprite is generally centered within 50 km of the parent +CG. Assuming the +CG provides the range, single-image photogrammetric analyses provide estimates of the maximum vertical extent of the sprites. For this storm the sprite tops averaged 77 km with a maximum of 88 km. The bases averaged 50 km but with a few sprite tendrils extending as low as 31 km. (44 refs.)

Rapid onset, rapid decay perturbations (RORDs) of subionospheric VLF propagation require highly localized or laterally structured plasma at low altitudes to explain the wide angle scattering observed and the rapid decay. Simultaneous occurrence of RORDs and red sprites, illustrated by a single event, together with VLF phase and group delay measurements from a pair of spaced receivers suggest that RORDs are produced by scattering from conducting columns at the position and with the lateral shape of the sprite. The sprite luminosity decays much faster than the RORDs which depend on the sprite conductivity and so plasma density. Plasma is also produced near the sprite plasma by energetic electrons precipitated from the magnetosphere by ducted whistlers and after the expected whistler and electron propagation delay. This whistler-induced electron precipitation (WEP) plasma produces wide angle VLF scattering similar to that by sprite plasma, implying similar lateral fine structure. This suggests that the processes leading to sprites also produce whistler ducts in the magnetosphere. (21 refs.)

We have determined the molecular internal energy distribution in the Nsub 2 Bsup 3 Pi sub g state from the fluorescence measured during the observations of sprites during 1995. Spectrally resolved data from two different instruments and three different sprites are compared with theoretical spectra to obtain excited state vibrational distributions. Energy dependent electron excitation cross-sections and laboratory data were used to estimate the energies of electrons producing the red sprite radiance. Implications for chemical production in the mesosphere and critical future measurements are discussed. (17 refs.)

Scattering by the conductive columns of red sprites of VLF waves ('VLF sprites') traveling in the Earth-ionosphere waveguide is characterized by wide angle scattering (to 180 degrees ). This enabled the first VLF detection of sprite conductivity and is now used routinely for sprite detection and location by measurement of the phase and amplitude at both of the frequencies transmitted by NAA, NSS, NLK and NPM. In a blind test using only one VLF receiver and the first three of these transmitters, all of the VLF events fitting the criteria for VLF sprites were found to correspond to optical sprites and virtually all of the sprites observed optically corresponded to VLF events. Using only a single receiver in the high noise environment of local thunderstorms enabled the range to the sprite to be found to within 100 km and the direction to within 90 degrees . Use of a large number ( 10) of suitably spaced antennae greatly increases the location accuracy and provides some information on the lateral structure of sprite conductivity. (14 refs.)

Red sprites occur high above the stratosphere, just under the ionosphere. Although the first reported observation was over 100 years ago, and the first theory was 40 years ago, only over the last year or so has the subject spread into the popular science magazines, and into the secular media. Most of the studies of the sprite structure have been optical, using the light they emit for a few tens of milliseconds for imaging (low-light video and photography) and spectroscopy. Here, we concentrate on the scattering by sprites of man-made VLF radio waves. This scattering shows that the columnar elements of sprites have a substantial electrical conductivity. (23 refs.)

New observations of sprites (cloud-ionosphere luminous discharges above thunderstorms) were made from the Yucca Ridge Field Station 20 km northeast of Fort Collins, Colorado, on the night of July 11-12, 1994, as part of a summer 1994 observing campaign. The sprites appeared above a moderate mesoscale convective complex mostly over Kansas at a range of about 270 km. The sprites were observed with both wide-field and telescopic image-intensified CCD TV cameras, a telescopic photometer system, and a 1- to 50-kHz band VLF sferics receiver. This paper is based on five 1-s data intervals containing bright sprites, smaller sprites, and cloud and sky flashes. Telescopic TV images of bright sprites had a fan-shaped upper plume with very fine features not well resolved by the TV, but dendritic (upward forked) and vertically striated forms adjacent to these plumes and bright points of luminosity around the plume-shaped regions. Many sprites consisted entirely of groups of vertically aligned striations which sometimes appeared to diverge from a common point of origin at cloud tops. All sprites in the present data sample were preceded by a cloud to ground (CG) stroke with a coincident sferic and sky flash. All CG strokes associated with sprites were positive, and most were 100 kA or more inferred peak current. From the photometer, the duration of the CG-induced sky flashes was about 3 ms and the additional sprite total light curve was also about 3 ms. The puzzling feature that the total duration of TV images of sprites was often longer than the photometric values is discussed and an explanation given. The sprites were attributed to strong negative charging, following the positive CG stroke, of a localized cloud top region which produced an intense electric field and a luminous discharge in the cloud-ionosphere region. The concept of 'break-even' electric fields suggested by McCarthy and Parks may explain discharge initiation with moderate field strengths. (13 refs.)

Fleeting columns of luminosity occurring above large thunderstorms at 50-90 km altitude, presently known as sprites, were imaged with an intensified video charge coupled device (CCD) camera during a July 1995 ground-based campaign near Fort Collins, Colorado. These unfiltered intensified images reveal detailed spatial structure within the sprite envelope. The temporal resolution of standard interlaced video imagery is limited by the 60 fields per second acquisition rate (16 ms). The specific CCD used, however, is subject to bright events leaking into the readout registers, allowing time-resolution on the order of the linescan rate (63 mu s). Typical sprite onset is found to follow the associated cloud lightning by 1.5 to 4 ms. The onsets of the individual sprites within a cluster are generally, but not always, simultaneous to within 1 ms. Sprites tend to have a bright localized core, less than 2 km in horizontal dimension, which rises to peak intensity within 0.3 ms and maintains this level for 5 to 10 ms before fading over an additional 10 ms. (12 refs.)

VLF perturbations on signals propagating along great-circle-paths (GCP) through electrically active midwest thunderstorms are associated with luminous high altitude glows (referred to as sprites) observed from aircraft or ground. The data constitutes the first evidence that the physical processes leading to sprites also alter the conductivity of the lower ionosphere. (19 refs.)

Large quasi-electrostatic (QE) fields above thunderclouds produce an upward traveling beam of approximately 1 MeV runaway electrons which may contribute to the production of optical emissions above thunderclouds referred to as red sprites. Results of a one dimensional computer simulation model suggest that the runaway electrons can produce optical emissions similar in intensity and spectra to those observed in red sprites, but only for large QE fields produced by positive cloud-to-ground discharges lowering 250 C or more to ground from an altitude of at least 10 km. Differences in predicted optical spectra from that of other mechanisms suggest that the runaway electron mechanism can be readily tested by high resolution spectral measurements of red sprites. (19 refs.)

In two summertime mesoscale convective systems (MCSs), mesospheric optical sprite phenomena were often coincident with both large-amplitude positive cloud-to-ground lightning and transient Schumann resonance excitations of the entire Earth-ionosphere cavity. These observations, together with earlier studies of MCS electrification, suggest that sprites are triggered when the rapid removal of large quantities of positive charge from an areally extensive charge layer stresses the mesosphere to dielectric breakdown. (47 refs.)

The intense plasma perturbations in the middle atmosphere, produced by the high-altitude discharges and appearing as luminescent columns ('red sprites') are considered. A model of the perturbations of the kind relaxing in the mesosphere after termination of the luminescent phase has been developed. Analysis solutions for the nonstationary continuity equations for electrons, positive and negative ions in the plasma columns have been obtained. It is shown that ion density perturbations should be taken into account in the analysis of VLF wave reflection from the plasma formations. The ion density perturbations also can increase the life time of 'sprite' like formations in the middle atmosphere. (12 refs.)

Telescopic images of sprites show a wide variety of generally vertical but also slanted fine structure, including branching tree-like shapes and well defined but isolated columns, with transverse spatial scales ranging from tens of meters to a few hundred meters at  60-85 km altitude. Simultaneous analysis of radio atmospheric and lightning data indicates that specific columnar regions are selectively excited by successive discharges. (16 refs.)

A global network of three electromagnetic measurement instruments is used to simultaneously record time series of globally observable extremely-low-frequency (ELF) magnetic field disturbances which propagate with little attenuation around the globe within the Earth-ionosphere cavity. The triangulation of individual lightning flashes results in a picture of the temporal evolution of intense lightning discharge occurrences on the planetary scale during April 1998. The lightning flash charge moments are calculated with the short pulse approximation of the normal mode expansion. The majority of the triangulated lightning discharges exhibit charge moments with a potential to excite mesospheric sprites and  5-20% may account for air breakdown at sprite altitudes in  50-70 km height. (21 refs.)

A physical picture and quantitative two-dimensional electromagnetic modeling of mesospheric electric field transients produced by cloud-to-ground (CG) lightning discharges with short duration currents (<0.5 ms) are presented. The range of applicability of existing quasi-electrostatic models of sprites and the physical conditions under which relatively weak CG lightning discharges (thundercloud charge moment changes less than 50 C *10 km) may initiate sprites are discussed in the context of experimental findings. (23 refs.)

We consider observations by Dowden et al. (1997) of very strong early/fast Trimpis or VLF sprites observed at Darwin, Australia, only 2,000 km from the US Navy transmitter, NWC (21 degrees 48'S, 114 degrees 9'E). The unperturbed signal strength was about 20 mV/m during the Trimpi observations (at night). The echo amplitudes of the strong Trimpis ranged up to 8 mV/m, some 30 dB stronger than those studied previously in Colorado. The much higher signal/noise ratio enabled accurate measurements of the time variation of the amplitude and phase of the echoes. (10 refs.)

A new and improved model of red sprites is presented. Emphasis is placed on accounting for the puzzling observation of the spatial structure in the emissions. The model relies on the electromagnetic pulse (EMP) fields created by a horizontal lightning discharge and includes the observed fractal structure of such discharges in the computation of the EMP power density. It is shown that the model can account for the observed spatial structure of the red sprites while reducing the typical charge required to approximately 100 C. (35 refs.)

A model of the 4.26 mu m infrared emission due to red sprites is presented. The model considers the generation of nitrogen `vibrons' due to the collisions of the nitrogen molecules with the electrons energized by the electric field from lightning, followed by the transition of the nitrogen vibrons to the COsub 2(001) vibrational level, with a lifetime much shorter than that of nitrogen. The infrared photons of wavelength lambda =4.26 mu m radiated by the COsub 2(001) propagate through the optically thick atmosphere; therefore, this emission could best be observed from space. The model computes the infrared radiance of sprites, as well as the energy collected by a state-of-the-art space infrared detector, and estimates the signal to background ratio. (23 refs.)

The results of a quasi-electrostatic electron heating model were combined with a time dependent Nsub 2 vibrational level population model to simulate the spectral distributions and absolute intensities observed in red sprites. The results include both Nsub 2 excited state vibrational level populations and time profiles of excited electronic state emission. Due to the long atmospheric paths associated with red sprite observations, atmospheric attenuation has a strong impact on the observed spectrum. We present model results showing the effect of atmospheric attenuation as a function of wavelength for various conditions relevant to sprite observations. In addition, our model results estimate the variation in the relative intensities of a number of specific Nsub 2 emissions in sprites (1PG, 2PG, and VK) in response to changes in observational geometry. A sprite spectrum, measured from the Wyoming Infrared Observatory (WIRO) on Jelm Mountain, during July, 1996, has been analyzed and includes Nsub 2 1PG bands down to nu '=1. In addition to Nsub 2 1PG, our analysis of this spectrum indicates the presence of spectral features which are attributable to Nsub 2sup + Meinel emission. However, the presence of Nsub 2sup +(Asup 2 Pi sub u) should be considered preliminary. The importance of both the populations of the lower levels of the Nsub 2(Bsup 3 Pi sub g) and the Nsub 2(Bsup 3 Pi sub g)/Nsub 2sup +(Asup 2 Pi sub g) population ratio in the diagnosis of the electron energies presented in red sprites is discussed. While the current spectral analysis yields a vibrational distribution of the Nsub 2(Bsup 3 Pi sub g) which requires an average electron energy of only 1-2 eV, model results do indicate that the populations of the lower levels of the Nsub 2(Bsup 3 Pi sub g) will increase with increases in the electron energy primarily due to cascade. (81 refs.)

The authors describe the terrestrial excitation of horizontal magnetic field variations in the Pc 1 frequency range (0.2-5.0 Hz) by tropospheric, sprite-associated lightning flashes, measured  1900 km west from the source. These variations, which they call ultra-slow tails, exhibit amplitudes on the order of tens of pT, they have a duration of  3 seconds, and they occur immediately following the initial pulse of the sprite-associated lightning flash. The ultra-slow tails exhibit two peaks in the frequency domain at 0.67 Hz and 1.67 Hz. The mean polarization ellipses at these two frequencies are oriented  45 degrees clockwise from geographic north and exhibit right-hand and left-hand polarization respectively with a weak ellipticity of  0.1. The horizontal magnetic, intensity of the initial pulse is related to the horizontal magnetic intensity of the ultra-slow tail, in agreement with the interpretation of ultra-slow tails as ionospheric Alfven resonances. (20 refs.)

The results of numerical calculations of intensity and spectra of optical emissions from red sprites produced by runaway air breakdown in the atmosphere are presented. It is shown that the optical emissions from red sprites consist of two components: (1) short-term (t approximately=0.3-2 ms) emissions produced as a result of dissipation of an energetic electron beam in air; (2) long-term (t approximately=2-10 ms) emissions produced by a population of low-energy electrons in an electric field. The long-term optical emissions are calculated for all low-energy electrons, including the secondary low-energy electrons produced by the relativistic electron beam, ambient background electrons, and electrons produced as a result of regular breakdown. The theoretical results are compared with observational data. (40 refs.)

VLF to MF backscatter from 'red sprites' is investigated by the use of a one-dimensional model, in which the sprite is considered as an infinite vertical slab of plasma. The reflection (backscatter) and transmission (forward-scatter) coefficients are found for a wide range of slab parameters. For a slab in which such parameters vary with altitude, the approximation is made that the scatter coefficients vary with altitude accordingly. A conductivity of 10sup -4 S/m is found sufficient to explain the observed magnitudes of backscatter (large angles up to 180 degrees ), provided this or higher conductivity prevails over a substantial part of the cloud-ionosphere altitude range. The effect of the geomagnetic field on the conductivity at altitudes up to 80 km, and wave frequencies up to 10 MHz, is found to be of little consequence. (12 refs.)

Quasi-electrostatic (QE) fields that temporarily exist at high altitudes following the sudden removal (e.g., by a lightning discharge) of thundercloud charge at low altitudes lead to ambient electron heating (up to  5 eV average energy), ionization of neutrals, and excitation of optical emissions in the mesosphere/lower ionosphere. Model calculations predict the possibility of significant (several orders of magnitude) modification of the lower ionospheric conductivity in the form of depletions of electron density due to dissociative attachment to Osub 2 molecules and/or in the form of enhancements of electron density due to breakdown ionization. Results indicate that the optical emission intensities of the 1st positive band of Nsub 2 corresponding to fast ( 1 ms) removal of 100-300 degrees C of thundercloud charge from 10 km altitude are in good agreement with observations of the upper part ('head' and 'hair' [Sentman et al., 1995]) of the sprites. The typical region of brightest optical emission has horizontal and vertical dimensions  10 km, centered at altitudes 70 km and is interpreted as the head of the sprite. The model also shows the formation of low intensity glow ('hair') above this region due to the excitation of optical emissions at altitudes  85 km during  500 mu s at the initial stage of the lightning discharge. Comparison of the optical emission intensities of the 1st and 2nd positive bands of Nsub 2, Meinel and 1st negative bands of Nsub 2sup + and 1st negative band of Osub 2sup + demonstrates that the 1st positive band of Nsub 2 is the dominating optical emission in the altitude range around  70 km, which accounts for the observed red color of sprites, in excellent agreement with recent spectroscopic observations of sprites. Results indicate that the optical emission levels are predominantly defined by the lightning discharge duration and the conductivity properties of the atmosphere/lower ionosphere (i.e., relaxation time of electric field in the conducting medium). The model demonstrates that for low ambient conductivities the lightning discharge duration can be significantly extended with no loss in production of optical emissions. The peak intensity of optical emissions is determined primarily by the value of the removed thundercloud charge and its altitude. The preexisting inhomogeneities in the mesospheric conductivity and the neutral density may contribute to the formation of a vertically striated fine structure of sprites and explain why sprites often repeatedly occur in the same place in the sky as well as their clustering. Comparison of the model results for different types of lightning discharges indicates that positive cloud to ground discharges lead to the largest electric fields and optical emissions at ionospheric altitudes since they are associated with the removal of larger amounts of charge from higher altitudes. (89 refs.)

Three different types of optical phenomena have been observed at high altitude above thunderstorms: an enhanced airglow ('elves') at roughly  90 km; a reddish glow ('sprites') from 50 to 90 km; and an upward moving, bluish emission ('jets') below 40 km. A likely explanation for some or all of these phenomena is gas breakdown caused by the electromagnetic fields of lightning discharges. This paper examines the connection between these fields and breakdown at high altitude, using both analytic models and numerical simulations. Included in the calculations are the radiation fields from the lightning return stroke and the quasi-static fields from the continuing current. The different nature of the two fields is shown to produce two distinct types of breakdown, with characteristics similar to those of elves and sprites. Also mentioned is a third breakdown mechanism which may account for blue jets. (23 refs.)

The discovery of electric discharges above thunderstorms has generated intense scientific interest. Studies of these `sprites' and `jets' have focused on their characterization by optical and radio techniques, with radar measurements of the causative storms providing insight into related weather. The authors describe a method for investigating the electric field structure above thunderstorms using ground-based radar to observe chaff dispersed by rockets. Slender conducting or dielectric chaff will generally align itself with the ambient electric field. This alignment is readily detected by appropriate configurations of polarimetric radar(s) such as are now used in meteorology to observe the nature and motion of hydrometeors. This is especially convenient as it permits the thunderstorms associated with sprites and jets to be characterized with the same experimental facility. This paper renders a preliminary examination of factors such experiments would entail and features that a multiparameter radar might utilize to probe chaff dispersed by small rockets. Monostatic and bistatic radar measurements of scatter from chaff provide a powerful tool to study electric fields associated with sprites and jets as well as other atmospheric electric fields. (29 refs.)

A TV slit spectrograph was used to obtain the first optical spectra of sprites. Twenty-five events were observed over a thunderstorm on the border of Nebraska and Colorado on the night of 22 June, 1995 between 0700 and 0900 UT. For 10 of these events optical spectra were measured in the wavelength range from 540 to 840 nm. After correcting for the spectrograph response function, digitized spectrograph video images are used to measure the wavelengths of and ratios between the emissions. All emissions are found to be of the first positive bands of Nsub 2. There is no evidence of the Meinel bands of Nsub 2sup + indicating that the mechanism responsible for sprites produces little or no ionization at 70 km altitude. (17 refs.)

Low light level monochrome television observations obtained from the ground and from the space Shuttle, and low light level color and monochrome television images obtained from aboard jet aircraft, have shown that intense lightning in mesoscale thunderstorm systems may excite at least two distinct types of optical emissions that together span the space between the tops of some thunderstorms and the ionosphere. The first of these emissions, dubbed 'sprites,' are luminous red structures that typically span the altitude range 60-90 km, often with faint bluish tendrils dangling below. A second, rarer, type of luminous emission are 'blue jets' that appear to spurt upward out of the anvil top in narrow cones to altitudes of 40-50 km at speeds of approximately 100 km/s. In the paper the principal observational characteristics of sprites and jets are presented, and several proposed production mechanisms are reviewed. (43 refs.)

The dual jet aircraft Sprites94 campaign yielded the first color imagery and unambiguously triangulated physical dimensions and heights of upper atmospheric optical emissions associated with thunderstorm systems. Low light level television images, in both color and in black and white, obtained during the campaign show that there are at least two distinctively different types of optical emissions spanning part or all of the distance between the anvil tops and the ionosphere. The first of these emissions, dubbed 'sprites' after their elusive nature, are luminous structures of brief (<16 ms) duration with a red main body that typically spans the altitude range 50-90 km, and possessing lateral dimensions of 5-30 km. Faint bluish tendrils often extend downward from the main body of sprites, occasionally appearing to reach cloud tops near 20 km. The principal characteristics of red sprites as observed during the Sprites94 campaign are described. (15 refs.)

Observational reports on high altitude flashes and their mysterious characteristics are critically examined. Results obtained from electromagnetic observations in conjunction with the intense thunderclouds are pointed out and the possible consequences are discussed. (27 refs.)

For original paper see Radio Sci., vol.33, no.6, p.1655-68 (1998). Nickolaenko comments on various aspects of the original paper, and suggests that that treatment will be valid for the HF range but not for the ELF-VLF range. (2 refs.)

It has been suggested that optical flashes observed in the upper atmosphere above giant thunderstorms (red sprites) are due to streamers. Such streamers are initiated in the lower ionosphere by electron patches caused by electromagnetic radiation from horizontal intracloud lightning and then develop downward in the static electric field due to the thundercloud. The triggering conditions of streamer development are analysed. Using similarity relations, known characteristics of streamer tips obtained earlier in laboratory conditions are extended to a description of streamers in rare air. Streamer growth in the nonuniform atmosphere is calculated. It is shown that streamers first appear at a height of about 80 km and then grow downward to slightly below 50 km, where they are terminated. This is in agreement with red sprite observations. An altitude distribution of the streamer generated plasma is obtained. The simple models of streamer development presented in this paper could be applied for computations of streamers growing in other conditions. (24 refs.)

Once dismissed as figments of pilots' imaginations, strange flashes appearing above thunderstorms have been confirmed as entirely new forms of lightning. Known as sprites, elves, blue jets and gamma-ray events, these high-altitude phenomena arise through a physics all their own. Their features and origin are discussed. (4 refs.)

Horizontal magnetic field variations in the frequency range of the Earth-ionosphere cavity resonances are observed at Silberborn, Germany, simultaneous with 19 sprite-associated lightning flashes in the midwestern United States, on July 15, 1995. The measured horizontal magnetic intensities are linearly related to the horizontal magnetic intensities of slow tails of radio atmospherics which were simultaneously recorded at Palmer Station, Antarctica. Enhancement of the Earth-ionosphere cavity resonances is verified by spectral analysis, and the measured arrival azimuths are in agreement with the expected orientation of the Poynting vector along the great-circle path. The estimated accuracy of the arrival azimuth determination is on the order of +or-5 degrees. (17 refs.)

Radio atmospherics launched by sprite-producing positive cloud-to-ground lightning flashes and observed at Palmer Station, Antarctica, exhibit large ELF slow tails following the initial VLF portion, indicating the presence of continuing currents in the source lightning flashes. One-to-one correlation of sferics with NLDN lightning data in both time and arrival azimuth, measured with an accuracy of +or-1 degrees at  12,000 km range, allows unambiguous identification of lightning flashes originating in the storm of interest. Slow-tail measurements at Palmer can potentially be used to measure continuing currents in lightning flashes over nearly half of the Earth's surface. (14 refs.)

Observations of optical phenomena at high altitude above thunderstorms using a multichannel high-speed photometer and image intensified CCD cameras were carried but at Yucca Ridge Field Station (40 degrees 40' N, 104 degrees 56' W), Colorado as part of the SPRITES'95 campaign from 15 June to August 6, 1995. These new measurements indicate that diffuse optical flashes with a duration of <1 ms and a horizontal scale of  100-300 km occur at 75-105 km altitude in the lower ionosphere just after the onset of cloud-to-ground lightning discharges, but preceding the onset of sprites. Here we designate these events as 'elves' to distinguish them from 'red sprites'. This finding is consistent with the production of diffuse optical emissions due to the heating of the lower ionosphere by electromagnetic pulses generated by lightning discharges as suggested by several authors. (22 refs.)

Quasi-electrostatic (QE) fields which exist above thunderclouds after lightning discharges can lead to the formation of columnar channels of breakdown ionization and carrot-like vertical luminous structures with typical transverse dimension  5-10 km spanning an altitude range from  80 km to well below  50 km. The carrot-like forms closely resemble those observed in sprites. Results indicate that the appearance of optical emissions can be significantly delayed in time ( 1-20 ms) with respect to the causative lightning discharge. (20 refs.)

Ultra low frequency magnetic field measurements made 500-2000 km from positive lightning discharges show a signature that is consistent with unusually high amplitude cloud-to-ground continuing lightning current. The magnitude of this nearly constant current moment is as large as 60 kA km and can last at half this amplitude for longer than 150 ms, thereby moving 640 C or more (assuming a 7 km vertical channel length) to the ground after the return stroke. This total charge transfer is more than an order of magnitude greater than most previously reported continuing currents in positive discharges. Three cases analyzed show this strong continuing current flows before, during, and after sprites that initiate more than 40 ms after the return stroke. Accounting for this continuing current, quantitative analysis shows that the total vertical lightning charge moment changes are large enough to produce mesospheric electrical breakdown and long-delayed sprites. (22 refs.)

An extension of an earlier paper (see ibid., vol.39, p.1051-4, July 1991) on this subject is hereby developed to allow for the generalization of the Born approximation to account for eddy currents in the column. Also, the relevance to VLF sprites is pointed out; these are ionized columns extending from the cloud tops to the low ionosphere. (5 refs.)

Detailed 2D hydrodynamic and quasi-electrostatic simulations of high-altitude discharges driven by runaway air breakdown are presented for four cases, corresponding to sprites initiated by positive cloud-to-ground lightning strikes in which 200 C of charge is neutralized at an altitude of 11.5 km in 10, 7, 5 and 3 ms. We find that the computed optical emissions agree well with low-light level camera images of sprites, both in terms of the overall intensity and spatial distribution of the emissions. Our results show the presence of blue emissions extending down to 40 km (blue tendrils) and red sprite tops extending from 50 to 77 km. Simulated spectra show that Nsub 2 1st positive emissions dominate in the wavelength range from 550 to 850 nm, in good agreement with observations. Strong radio pulses with durations of  300 mu s and peak electric field amplitudes ranging from 20 to 75 V/m at an altitude of 80 km and an approximate distance from the discharge of 50 km were computed. The magnitude and duration of these pulses is sufficient to cause breakdown and heating of the lower ionosphere (80-95 km) and leads us to suggest that sprites may also launch the EMP responsible for the production of elves. The computed values for the gamma -ray fluxes are in agreement with observations of gamma -ray bursts of atmospheric origin and the peak secondary electron densities which we obtain are in good agreement with HF echoes at mesospheric heights and associated with lightning. (59 refs.)

This review considers the different models that have been developed to explain a class of phenomena that occur above lightning storms. These phenomena have been named elves, red sprites and blue jets. The elves appear between 90 and 70 km altitude and extend over several 100 km horizontally. They are visible for less than 0.1 ms. Red sprites cover a range of altitudes from 80 to 55 km with narrow tendrils extending below 55 km. Horizontally they are 20-30 km wide. Their visible lifetime is from a few to some tens of ms. Blue jets propagate from cloud tops (15 km) to an altitude of 40 km with a velocity of 100 km/s which gives a lifetime of 300 ms. In all of the models, the energy source is the electric fields associated with the lightning-the quasistatic fields due to the original charge distribution, the electromagnetic pulse due to the propagation of the return stroke or the quasistatic fields due to the charge redistribution by the currents. There are two different models to explain the heating of the neutral atmosphere by these electric fields. These models accelerate either the ambient thermal electrons (<eV) or high energy, cosmic-ray-generated MeV electrons. These electrons in turn collisionally heat the neutrals and produce the heating, ionization and optical emissions. (47 refs.)

Measurements of ELF/VLF radio atmospherics (sterics) at Palmer Station, Antarctica, provide evidence of active thunderstorms near the inferred source regions of two different gamma-ray bursts of terrestrial origin (Fishman et al., 1994). In one case, a relatively intense steric occurring within +or-1.5 ms of the time of the gamma-ray burst provides the first indication of a direct association of this burst with a lightning discharge. This steric and many others launched by positive cloud-to-ground (CG) discharges and observed at Palmer during the periods studied exhibit `slow tail' waveforms, indicative of continuing currents in the causative lightning discharges. The slow tails of these sterics are similar to those of sterics originating in positive CG discharges that are associated with sprites. (18 refs.)

Imagery and spectra of high altitude luminous flashes, otherwise known as sprites, occurring in the stratosphere/mesosphere above electrically active cumulonimbus clouds were acquired on July 16, 1995 from an observation site near Ft. Collins, Colorado. The spectra, resolved from approximately 4500-8000 AA included four spectral features in the 6000-7600 AA region which have been identified as the Nsub 2 1 PG system with Delta v=2, 3, and 4 from the v=2, 4, 5, 6 vibrational levels of the Bsup 3 pi sub g state. The spectra were lacking in other features such as the Nsub 2sup + Meinel or the Nsub 2sup + 1st neg system indicating that the electron energy causing the excitation is quite low. (17 refs.)

A balloon sounding of electric field in the trailing stratiform cloud of a bow echo mesoscale convective system reveals only two substantial in-cloud positive charge regions. These charge regions are located at altitudes of 5.1-5.6 km and 6.4-6.8 km, above the level of 0 degrees C at 4.2 km. The two positive charge regions are the likely sources of six positive cloud-to-ground flashes with large peak currents (>32 kA) that occurred within 60 km of the balloon during its flight. The amount of charge transferred by three of these positive flashes that made Q bursts is calculated in the range of 97-196 C. Flashes of this sort are known to produce sprites and elves in the mesosphere. The positive charge regions in this stratiform cloud are substantially lower than the 10-km altitude commonly assumed for the positive charge in many sprite modeling studies. (28 refs.)

Very strong phase and amplitude perturbations (Trimpis) of the very strong VLF transmission from NWC received in Darwin, Australia, enabled accurate measurement of the amplitude and phase of the scattered signal and their time variation. The amplitude of the scattered signal decays as the logarithm of time, quite at odds with the exponential decay observed on classic Trimpis. During the amplitude decay, the phase of the scattered signal decreased at a decreasing rate. This is shown to be consistent with scattering from a bundle of sprite-like, conducting columns extending some 50 km below the base of the ionosphere. (11 refs.)

A mechanism for the nonlinear excitation of the ionospheric Alfven resonator by elves- or/and sprites-produced lightning discharge is proposed. The source of the time-varying nonlinear current, located at altitudes below 95 km, is due to the large impulse electron heating and breakdown of the atmosphere by a strong tropospheric discharge. The discussed mechanism may be responsible for anomalously large ULF events observed onboard the satellites above atmospheric weather systems. (26 refs.)

The origins of mysterious gamma -ray and radio flashes recently detected by satellite-based instruments passing over thunderstorms are examined in the context of upward propagating discharges initiated by runaway air breakdown. Preliminary calculations normalized by the recent optical measurements of so-called sprites indicate that the runaway mechanism may well be the source of these emissions. If this is true, then upward discharges represent the first known manifestation of a fundamental, new process in plasma physics. (50 refs.)

gamma -ray flashes of atmospheric origin as well as blue jets and red sprites are naturally explained by high-altitude discharges produced by runaway air breakdown. The authors present the first detailed model of the development of upward propagating discharges and compute optical and gamma -ray emissions that are in excellent agreement with observations. According to their theory, such discharges represent the first known manifestation of runaway air breakdown, a fundamental new process in plasma physics. (26 refs.)

Quasi-electrostatic fields that temporarily exist at high altitudes following the sudden removal (e.g. by a lightning discharge) of thundercloud charge at low altitudes are found to significantly heat mesospheric electrons and produce ionization and light. The intensity, spatial extent, duration and spectra of optical emissions produced are consistent with the observed features of the red sprite type of upward discharges. (19 refs.)

A three-dimensional Monte Carlo model of the uniform relativistic runaway electron breakdown in air in the presence of static electric and magnetic fields is used to calculate electron distribution functions, avalanche rates, and the direction and velocity of avalanche propagation. The authors also derive the conditions required for an electron with a given momentum to start an avalanche in the absence of a magnetic field. The results are compared to previously developed kinetic and analytical models and their own analytical estimates, and it is concluded that the rates used in many early models [e.g., Lehtinen et al., 1997; Taranenko and Roussel-Dupre, 1996; Yukhimuk et al., 1998; Roussel-Dupre et al., 1998] are overestimated by a factor of  10. The Monte Carlo simulation results are applied to a fluid model of runaway electron beams in the middle atmosphere accelerated by quasi-electrostatic fields following a positive lightning stroke. In particular, the authors consider the case of lightning discharges which drain positive charge from remote regions of a laterally extensive (>100 km) thundercloud, using a Cartesian two-dimensional model. The resulting optical emission intensities in red sprites associated with the runaway electrons are found to be negligible compared to the emissions from thermal electrons heated in the conventional type of breakdown. The calculated gamma ray flux is of the same order as the terrestrial gamma ray flashes observed by the Burst and Transient Source Experiment detector on the Compton Gamma Ray Observatory. (32 refs.)

A spatial distribution and spectra of optical emissions of gigantic upward atmospheric discharges governed by a relativistic runaway electron avalanche were calculated. High-energy electrons of the avalanche excite emissions accounting for blue jets, which are observed immediately above thunderclouds, and produce a copious amount of secondary low-energy electrons. Some energy obtained by the secondaries from the field, is deposited in optical emissions, which was shown to account for red sprites observed at higher altitudes. Calculated distribution of nitrogen molecules over vibration states Bsup 3 Pi sub g(v) agrees with experimental data. (24 refs.)

Smoke from forest fires in southern Mexico was advected into the U.S. southern plains from April to June 1998. Cloud-to-ground lightning (CG) flash data from the National Lightning Detection Network matched against satellite-mapped aerosol plumes imply that thunderstorms forming in smoke-contaminated air masses generated large amounts of lightning with positive polarity (+CGs). During 2 months, nearly half a million flashes in the southern plains exhibited