Abstract
An examination and analysis of video images of lightning, captured by the payload bay TV cameras of the space shuttle, provided a variety of examples of lightning in the stratosphere above thunderstorms. These images were obtained on several recent shuttle flights while conducting the mesoscale lightning experiment (MLE). The images of stratospheric lightning illustrate the variety of filamentary and broad vertical discharges in the stratosphere that may accompany a lightning flash. A typical event is imaged as a single or multiple filament extending 30 to 40 km above a thunderstorm that is illuminated by a series of lightning strokes. Examples are found in temperate and tropical areas, over the oceans and the land.
Introduction
Prior to 1990, the literature contained occasional reports of luminous flashes that extended to great heights in the stratosphere above thunderstorms. Boys (1926) believed such an event was a once in a lifetime observation. However, Wilson (1956) later speculated that a discharge between the top of a cloud and the ionosphere might often be the normal accompaniment of a lightning discharge to ground. Vonnegut (1980) and Vaughan and Vonnegut (1989) collected and published a number of eyewitness accounts of vertical lightning observations above thunderstorms. Although most of those reports could be taken as descriptions of lightning channels that terminated in the clear air rather than in the cloud, there were exceptional cases that differed from the description of a typical air discharge.
Interest in this phenomena increased considerably after Franz et al. (1990) obtained the first recorded observation of a stratospheric flash using a low light level video camera during the SKYFLASH program. They acquired a video image of two vertical plume discharges simultaneously extending into the clear sky above a thunderstorm. Vaughan et al. (1992) described a stratospheric flash captured in nighttime video images from the space shuttle that extended 31 km above the cloud tops. More recently, observations from the ground [Lyons and Williams, 1993; Lyons, 1994a; Lyons, 1994b] and from airborne platforms [Sentman and Wescott, 1994] have extended the observational data base of these stratospheric discharges. In addition, other recently reported phenomena may be closely coupled to the occurrence of stratospheric lightning. These include the detection of gamma-ray burst of atmospheric origin [Fishman et al., 1994], lightning-induced brightening of the air glow layer [Boeck et al., 1992] and unusual paired radio burst detected by the ALEXIS satellite [Holden et al., 1994].
An examination of video images of lightning, captured by the payload bay TV cameras of the space shuttle, has provided a variety of examples of luminous discharges in the stratosphere above thunderstorms during the period 1989 to 1991. In this paper we describe these observations. The images illustrate the variety of filamentary and broad vertical discharges in the stratosphere that may accompany a lightning flash. A typical event is imaged as a single or multiple filament extending 30 to 40 km above a thunderstorm that is illuminated by a series of lightning strokes.
Shuttle Video
The video images of stratospheric lightning were obtained from tens of hours of nighttime video collected from several recent shuttle flights while conducting the MLE [Vonnegut et al., 1985]. MLE used existing space shuttle cameras and transmission facilities on a nonpriority basis to view the Earth and thunderstorms at night. Lightning, when present, illuminates the storm from within by a multiple scattering process. When viewed from above, a lightning flash appears as a quasi-circular pool of light in the cirrus anvil cloud above the storm (storms without a significant anvil may exhibit a variety of illumination textures during a flash). In the shuttle data, most lightning storms were observed in oblique view because they were far from the ground track of the shuttle. The circular pool of cloud top light becomes football; shaped in oblique view. The term lightning or cloud flash; will be used in this paper to describe a cloud mass that is illuminated from within by an electrical discharge. These flashes include both intracloud and cloud-to-ground discharges. The lightning channels associated with these flashes, whether within or below the cloud, are not seen from space.
During initial analysis, we occasionally observed eye-catching transients in oblique views of lightning during video playback. Upon replay and freeze-frame examination of these events, luminous discharges (i.e., stratospheric lightning) were found in the clear air above the clouds. Most of the video examples captured from the viewpoint of space have the appearance of thin luminous filaments of glowing gases stretching into the stratosphere above a thunderstorm. Most images of stratospheric lightning also exhibit the horizon as a discontinuity in the image highlighted by the luminosity of a band of the airglow layer extending from the horizon to about 100 km. The initial reports of stratospheric lightning from shuttle video [Boeck et al., 1991; Vaughan et al., 1992] reported events that were viewed against a backdrop of stars and airglow. Using the experience of examining a number of cases set against a star background, this report also includes cases where there is a portion of Earth surface in the background of the flash. We believe that these latter cases are the same phenomena that were imaged at the limb of the Earth and cannot be dismissed as a projection of an extensive horizontal lightning flash at the Earth's surface.
Observations
We have identified 17 stratospheric discharges in the tens of hours of lightning video obtained during 1989-1991. Table 1 summaries a variety of statistics associated with each stratospheric flash. In Table 1, Time indicates the start time in UTC of the cloud flash associated with the stratospheric discharge. The UTC time of the event is obtained from a time window that is added in the TV studio. The time may vary from UTC by ±1 s. Start Delay is the delay in seconds in the appearance of the stratospheric flash following the start of the cloud flash. All the times given in Table 1 are determined to a resolution of a video field or 1/60 s (0.017 s). Two video fields constitute a standard video frame. Since the video integrate the input over the time of one video field, the TV images lack the time resolution to distinguish multiple lightning strokes occurring within the field. Hence it is not possible to determine the number of strokes in a flash.
Also in Table 1, the stratospheric flashes have been grouped by appearance into four broad classes: single filaments (SF), double filaments (DF), columnar discharges (C), and unclassified events (U). In addition, we have noted whether the discharge is continuous or displays a single break (SB) or multiple breaks (MB) along its length.
Finally, Table 1 provides estimates of the lightning flash rate of each storm producing a stratospheric discharge. In some instances the storm producing the stratospheric flash only produced a single discharge. This has been designated in the Flash Rate column as S. In a few cases the storm flash rate was not determined (ND). This usually occurred when it was not possible to separate the storm from neighboring storms due to spacecraft motion and camera operations (e.g., pan motions and zoom changes), or because the storm was very distant on the horizon. On a moonless night there was often no ground reference (e.g., city lights) or cloud reference information available to track the location of a particular storm.
Figure 1 provides a single captured image from each stratospheric flash event summarized in Table 1. Unfortunately, the transient visual impression is often not accurately conveyed by the captured image of a single video field since each event dynamically appears against a background of Earth or sky with accompanying electronic noise or "snow" in the video frame. Human vision systems seem to disregard most of the electronic noise. This may be analogous to viewing a sporting event in a snow storm.
Brief Description of Events
October 21, 1989 (Figs. 1a, 1b)
The first stratospheric flash was observed on this day at 10:34:20 UTC near the horizon as two distinct filaments (Fig 1a). The filaments contained multiple breaks or dark bands. The dimmest section was near the cloud top (i.e., near bottom of the filaments). This stratospheric flash accompanies an extremely bright cloud flash that produced a reflection on the vertical stabilizer of the shuttle (vertical stabilizer not shown in this cropped image). The flash rate of the specific storm that produced this event was not determined since movement of the camera and the lack of visible ground reference made it impossible to track the storm. However, the flash rate was definitely low.
A second stratospheric flash on this day (12:10:13 UTC) consisted of a single filament flash that extended upward over several video frames (Fig. 1b). The general appearance is similar to the plume of a rocket launch emerging from the illuminated cloud top. Other stratospheric flashes observed by the shuttle attained their full vertical development within the time of one video frame (0.017 s). Otherwise, in appearance, this flash appeared similar to the other single filament discharges observed. The storm in which this flash occurred appeared to be a supercell storm with a thick anvil that obscured the illuminated cloud except along the side and at a location presumably an overshooting turret. The storm complex in the vicinity of this event had a lightning flash rate greater than 50 flashes per minute. No other example in this paper was associated with such a high flash rate storm.
January 14-18, 1990 (Figs. 1c, 1d, 1e)
A horizontally extensive cloud flash that involved two or three cells initiated the double filament flash (21:12:35 UTC) shown in Figure 1c. Although it is not discernible in this image, each filament fanned out at the top. Note that the horizon and the atmospheric air glow layer are clearly visible in the figure. The general view was of two active thunderstorm complexes near the coast of West Africa. The particular location was at the edge of one of these thunderstorm areas. However, the storm responsible for this stratospheric flash appeared to produce only a single lightning discharge while it was in the field-of-view.
Figure 1d illustrates a stratospheric flash (18:53:25 UTC) that we have classified as a single column because of its greater width (as compared to a filament). The intensity of the column fluctuated during the 0.183 s (i.e., 5.5 frames) after its first appearance, with the upper portions remaining brighter than the lower portions of the column. Intensity fluctuations were often seen in the stratospheric flashes observed in the shuttle video, perhaps corresponding to multiple strokes not resolvable by the slow video frame rate. The relatively dark object in this image is the vertical stabilizer of the shuttle.
Figure 1e shows one filament of a double filament flash that occurred over Arkansas. The entire filament projected above the horizon. City lights from the east coast of the United States are visible in this image. The entire filament projects above the horizon.
April 26-28, 1990 (Figs. 1f, 1g)
Another example of a columnar discharge is shown in Figure 1g. This figure clearly shows a distinct bulge or blob of illumination at the upper end of the stratospheric flash. This perception of an upper bulge is further enhanced in this and other cases by the break in the illumination found just below the upper bulge. The storm associated with this flash was located at about 7.5° N, 4° E over the Gulf of Guinea. The column extended beyond the Earth limb against a star background. Several stars were identified in the scene. Using the angular separation of the stars to scale the distances in the scene, the length of the vertical column was estimated to be approximately 31 km [Vaughan et al., 1992].
October 6-8, 1990 (Figs. 1h, 1i)
Figure 1h shows a single filament stratospheric flash, with a single break in the illumination, and a bulge of illumination at the top observed on October 6, 1990, at 23:37:06 UTC. The flash extended above the horizon. The camera was viewing central Africa but there were no cities or other evidence to estimate the size of the storm. The location of interest was in view for 2 min and 33 s. In the field-of-view there were several active thunderstorm cells in a large cloud bank that was brightly illuminated by moonlight.
Figure 1i shows one of two single filament stratospheric flashes (23:43:31 UTC) that were obtained while the shuttle was passing over a large complex of thunderstorms in west Africa on October 8, 1990. The thunderstorm complex first came into view at 23:41:45 UTC. The location of interest was visible for 3 min and only a few flashes were seen near there. The camera had been exposed to extreme levels of light and several other images were burnt into the screen.
August 6-7, 1991 (Figs. 1j, 1k, 1l, 1m, 1n, 1o)
A total of six stratospheric flashes were observed over South America during the shuttle mission STS-43 on August 6-7, 1991. Unfortunately, the video quality was also the poorest during this mission as well. The stratospheric flash in Figure 1j (01:29:48 UTC) illustrates very clearly a fan-shaped structure at the top of the filament. We do not believe that this structure was an artifact of the video transmission problems that are also apparent in the image (note the alternating light and dark bars in images from this mission). Two min following this event, a double filament stratospheric flash was observed (Fig. 1k, 01:31:32 UTC).
The stratospheric flash shown in Figure 1l appeared to be a single flash reaching most of the way to the air glow layer. It was associated with a cloud source that was either blocked by intervening clouds near the horizon or was largely over the horizon. This visual blockage is believed to explain why the event apparently occurred simultaneously with the visible cloud flash (i.e., zero start delay).
Figures 1m, 1n, and 1o show the three remaining stratospheric flashes that occurred during this mission. In each of these events the bulge of illuminosity near the top of the flash was a prominent feature. In fact, the stratospheric flash shown in Figure 1o consisted only of the bulge of illumination occurring at approximately the same altitude as seen at other occasions. In Figure 1o there is no visible connection between the cloud and the luminosity above the storm.
September 17, 1991 (Fig. 1p)
The stratospheric flash shown in Figure 1p occurred in an area of moderately active thunderstorms. However, because of the dark background scene and occasional camera movement, it was not possible to determine the relative location and flash rate of the surrounding storms or the dimensions of the stratospheric flash.
November 27, 1991 (Fig. 1q)
A single filament flash was observed on a moonless night from an isolated storm over the Pacific ocean (Fig. 1q). The horizon and airglow are hidden in this high contrast picture with only a horizontal band of noise pulses to mark the approximate location of the airglow layer above the horizon. A single lightning flash occurred while this storm was in the field-of-view of the shuttle. Although the observation time of the storm could not be precisely determined for this case because of the lack of light, the flash rate of this storm can be estimated at less than 1 flash per minute.
Summary
Appearance
A typical event is imaged as a single or multiple filament extending 30 to 40 km above a thunderstorm that is illuminated by a series of lightning strokes. The stratospheric flashes are much less bright than the associated cloud flashes and seldom saturate the video image. Qualitatively, the brightness ranged from being just barely discernible above the video background level to much brighter than the airglow layer. Also, the upper portions of the stratospheric flash often tended to be brighter than the lower portions. Single and multiple breaks in the illumination were a commonly observed feature (72%) in these events. Many of the stratospheric flashes displayed an obvious broadening or bulge at the top of the flash. In a couple of cases (e.g., see Fig. 1j), the illumination appeared to flatten out like a pinhead at the top. In one example (Fig. 1o), only the upper illumination was observed with no connecting filament or channel being detected. Boeck et al. (1992) have previously reported a brightening in the airglow layer associated with a cloud flash.
Timing and Development
The mean duration of these events is found to be 0.13 s (i.e., 3.8 video frames). The average delay between the start of cloud illumination and the stratospheric flash is 0.358 s. The full vertical extent of the flash is established within the time resolution of one video field (17 ms). Most of the examples have maximum intensity at the start of the stratospheric flash followed by a continuous decay of luminosity. The observed decay time is determined by camera characteristics for any singular event. In a few cases the flash has several subsequent intensity maxima. This behavior suggests that some phenomena similar to multiple lightning strokes may be occurring.
It has not been possible to establish the direction of propagation of the stratospheric flashes observed in the shuttle video. The observation shown in Figure 1o of luminosity not connected to the thunderstorm luminosity that may be evidence of a downward direction of propagation. On the other hand, the stratospheric flash shown in Figure 1b is clearly an event that develops progressively upward during several frames and may indicate an upper direction of propagation. This event was also the only example associated with a high flash rate storm.
Dimensions
The length of a stratospheric flash can be estimated in those cases when the stars in the field-of-view can be identified. An accurate knowledge of the orbit parameters and the time of the event provides the position (including altitude) of the shuttle while the identification of the star field in the video image establishes the direction to the stratospheric flash. Using this information, the geographical location of the storm and its distance from the shuttle can be determined. The known angular separation of a star pair visible in the image near to the storm calibrates the image and provides the angular length of the stratospheric flash. The actual length of the flash is found from the previously computed range.
Figure 2 presents the photo analysis of the event obtained on October 6, 1990 (Fig. 1h). The analysis leads to the conclusion that the top of the stratospheric flash was 47 km above an assumed cloud top of 18 km. The combined height of the flash and cloud is 65 km above the Earth. Five shuttle examples have been analyzed in a similar fashion, resulting in estimates of stratospheric flash heights of 60 to 75 km above the Earth. Uncertainty arises from the rather coarse resolution of the digitized image and from not knowing the cloud top height or the location at which the stratospheric flash intersects the cloud top. For a typical case of an event near the horizon, the error in range in about 50 km which corresponds to a 2 to 3 km error in stratospheric length.
The width of the luminosity seems to vary considerably between events. Some examples show a very thin or even several thin vertical filaments while others appear as broad columns that may be some kilometers across. In addition, many events contain a bulge of illumination at the top of the flash that have dimensions on the order of kilometers. There is insufficient information to determine whether there are distinct types of stratospheric flashes or whether the shuttle observations merely provide an indication of a wide range of natural variation.
Cloud Flash
A stratospheric discharge always occurred in association with a cloud flash. In addition, this cloud flash was always the brightest and largest (i.e., in areal extent) illumination produced by the storm while it was in the field-of-view. The cloud flash typically preceded the stratospheric flash by a quarter to about a half second. In every case, the cloud flash continued after the stratospheric flash had disappeared, typically for a large fraction of a second. The mean duration of the cloud flashes in this study was 1.06 s. Of course, it should be noted that we would not recognize a stratospheric flash in the shuttle video if it was not associated with a cloud flash.
Storm Flash Rate
The typical stratospheric flash accompanies a cloud flash in a thunderstorm cell that exhibits a low to moderate flash rate. Table 1 shows that seven of the events were associated with storms that had flash rates less than about two flashes per minute and four events were associated with storms that had flash rates between three to ten flashes per minute. However, there was one example in which a stratospheric flash was produced in a storm complex that had a flash rate greater than 50 flashes per minute. The movement of the shuttle limits the observation time of any location to 2 to 3 min at most. In the four examples in Table 1 in which only a single discharge was observed, a storm flash rate less than 0.5 per minute can be inferred. As noted previously, in a few cases it was only possible to rule out high storm flash rates.
Geographical Distribution
The geographic distribution of events is biased by the limited opportunities to observe them. The major limitations are the number and duration of shuttle missions, the low inclination of the orbits, and the inability to observe these phenomena during daylight. Table 2 shows the approximate location of the stratospheric flash and the location of orbiter for each observation. The examples include Northern and Southern hemisphere cases, temperate and tropical areas, and oceanic and continental storms. Thus it appears that the conditions for these stratospheric discharges may occur over most regions of the globe.
Frequency of Occurrence
A rough estimate of the frequency of occurrence of stratospheric lightning can be made from the MLE observations gathered for 2 yr starting October 1989. This estimate is biased by the impression that the vertical discharge could not be identified against a bright background. Consequently, all cases that have been observed appear near the horizon. We estimate that vertical flashes can be identified with a frequency of the order of one in five thousand total lightning flashes seen from the space shuttle.
Conclusions
The interpretation of these images is made difficult by the lack of diagnostic electrical data. The most important parameter would be the current magnitude and the current waveform. The delay of a fraction of a second after the beginning of cloud illumination indicates that this phenomena is not time-coincident with initial development and propagation of the lightning channel. We speculate that this discharge is a secondary breakdown induced by the electric field produced by the displacement current component of a lightning discharge coupling into the global electric circuit. Perhaps an external ionizing event coincident with the temporary existence of an unrelaxed electric field in the stratosphere is needed to create the electrons that produce the visible discharge.
References
Boeck, W. L., O. H. Vaughan, Jr., R. J. Blakeslee, B. Vonnegut, M. Brook, and J. McKune, Jr., Lightning to the upper atmosphere: a vertical light pulse from the top of a thunderstorm as seen by a payload bay TV camera of a Space Shuttle, Proc. Int. Aerospace Lightning Conf., NASA Conf. Pub. 10,058, April 16-19, 1991.
Boeck, W. L., O. H. Vaughan, Jr., R. J Blakeslee, B. Vonnegut, M. Brook, Lightning induced brightening in the airglow layer, Geophy. Res. Lett., 19, 99-102, 1992.
Boys, C. V., Progressive lightning, Nature, 118, 749-750, 1926.
Fisher, J. R., Upward discharges above thunderstorms, Weather, 45, 451-452, 1990.
Franz, R. C., R. J. Nemzek, and J. R. Winckler, Television image of a large upward electrical discharge above a thunderstorm, Science, 249, 48-51, 1990.
Vaughan, O. H., Jr., R. J. Blakeslee, W. L. Boeck, B. Vonnegut, M. Brook, and J. McKune, Jr., A cloud-to-space lightning as recorded by the space shuttle payload-bay TV cameras, Mon. Wea. Rev., 120, 1459-1461, 1992.
Vaughan, O. H., Jr., and B. Vonnegut, Recent observations of lightning discharges from the top of a thunderstorm into the clear air above, J. Geophys. Res., 94, 131,179-131,182, 1989.
Vonnegut, B., O. H. Vaughan, Jr., M. Brook, P. Krehbiel, Mesoscale observations of lightning from space shuttle, Bul. Am. Met. Soc., 66, 20-29, 1985.
____________
Copyright 1994 by the American Geophysical Union.
Paper number 014
BOECK ET AL.: OBSERVATIONS OF LIGHTNING
Figure 1.