II. OBSERVATION EFFORTS AT CREIGHTON UNIVERSITY
Atmospheric Science Department in downtown Omaha. This location offered a good view
of potential thunderstorms from the Northwest , through the North, and through the East.
However, due to light effects from the greater Omaha Metropolitan area, we were unable
to adequately visualize distant thunderstorm imagery - we would be able to see only the
most brilliant flashes from C-I events. Later, a series of new sites was chosen. Our intent
was to find a location that had a ready electric power supply, as little city lighting as
available, and decent elevation, such as on a hill or tall building. I will continue with a
summary of the details surrounding our efforts.
A. Equipment used
Laboratory for the purpose of recording C-I imagery. These cameras have a rotating
"filter wheel" - a wheel containing several different opaque optical filters. Each filter is
designed to only detect electromagnetic radiation at a certain wavelength. The primary
camera has six optical filters, with spectral characteristics as shown in Table 2. The basic
camera can be seen in figure 3. Our primary efforts focused around filter (or channel)
number 5, at 7774 angstroms - one of the atomic oxygen lines. We chose this channel
because it provided the best looking imagery when compared to the other five filters and
their respective imagery. The term "best looking" should be interpreted to mean best gray
shade contrast - the camera will eventually produce a digitized image from
1. . . . . . . . N2+ . . . . .
. . . 4278 (+/- 1.87%)
2 . . . . . . . .HI . . . . . . . . 4861 (+/- 1.87%)
3 . . . . . . . .OI . . . . . . . . . 5571 (+/- 1.87%)
4 . . . . . . . .OI . . . . . . . . . 6300 (+/- 1.87%)
5 . . . . . . . .OI . . . . . . . . . 7774 (+/- 1.87%)
6 . . . . . . . .OH . . . . . . . . .7820 (+/- 1.87%)
the video input in an 8-bit graphic image (256 gray shades). Each of these images is the
exact same resolution - 756 (Horizontal) pixels by 485 (Vertical) pixels, so the only real
issue was the determination of adequate contrast. Channel 5 won easily: after trying to
image an active thunderstorm in late May 96, we noted that channels 1,2,3,4, and 6 were
not able to provide intelligible imagery. Thus, the decision was made to use this channel
on a consistent basis - in order to have some form of "baselined" imagery.
The VCR recorded the imagery in standard VHS format, for post-analysis. Attempts were
made to record rough geographic and navigational details. With the camera mounted on a
tripod, the camera field of view (FOV) , azimuth (with respect magnetic north), elevation,
and rough position were recorded. Geographic notes were taken of surrounding
landmarks for future reference. Also, notes were taken on the time. No attempt was made
to correct the camera's internal clock, but a time difference of 7 minutes, 32 seconds was
noted (the system time, for example, may say 04:07:32 GMT when it is really 04:00:00
GMT). The primary difficulty we experienced lay in the initial set-up of the entire
camera/TV/VCR, which required some re-engineering.
slower manner, frame by frame. Our ability to make observations of the recorded sprite
imagery was improved greatly.
B. Method of Employment
lighting conditions in downtown Omaha, our vantage point had limited utility. We set out
to find a series of different locations that would provide a better view with few
background lights from towns and cities, and a wide azimuthal look angle. One such
place was found in Bellevue, Nebraska, just south of Omaha. Our efforts consisted of
carrying the camera, and associated equipment from Creighton to the Bellevue Sprite
Observation Site (BSOS) , where we set up the camera on a hill located near 41o09'30"N
and 095o57'20"W (position only accurate to within +/- 10") at over 1110 feet above
mean sea level. At this location, we had a good view from the south-southwest, through
the west, to the north-northwest, with comparatively less city light interference from the
surrounding country towns. See Figure 4 for a view of the site. Our view to the west
southwest (azimuth: 240o magnetic) was especially good- there were no significant lights
to cause concern - just some repetitive aviation beacons, which were relatively low on the
horizon and easy to distinguish from other lights.
discussion of these events, just an overview of two nights in particular: the 15th and 22nd
of August, 1996. I will discuss these two because they were clearly the most prolific
nights all summer. Of the two nights, the 15th was by far the most remarkable display of
all, thus some care will be taken with this night.
1. Multiple events from August 15, 1996
a) Synoptic Weather Scenario
On 15 August 1996, 0600 GMT (Greenwich Mean Time), a low pressure system
was located over northern Michigan, with a strong cold front extending through Illinois,
northern Missouri, and southern Nebraska. In advance of this front, a series of
thunderstorms were developing. Cells were observed from Missouri, and thence into
southern Nebraska and Kansas (see figure 5 - provided by the National Weather Service) .
By 0400Z (0000 Eastern Time), a vigorous group of cells had started in the eastern
Colorado, and another group was losing strength in central Kansas. This is seen in the 0235Z
Radar Summary [figure 6] from the National Weather Service. and 7).
The cells that passed through the BSOS were all but dissipated by this time.
b) Location / Description of local conditions
hours prior (see Figure 6) . The area had rapidly cleared out, and there were stars in the
sky down to the horizon. No obscuration was evident at all. Noticing this improvement in
local visibility, we started reviewing radar summaries and satellite imagery. In addition,
we obtained lightning summaries from the National Lightning Detection Network
(NLDN) courtesy of the Air Force Global Weather Center (AFGWC - Now referred
to as the HQ Air Force Weather Agency). AFGWC also provided us with a look at imagery
from the Defense Meteorological Satellite Program - their Operational Linescan (OLS)
system can see lightning flashes at night. This gave us an excellent feel for the location of
actual lightning. Unfortunately, due to Department od Defense data restrictions, we are unable
to show that imagery here on this WWW site currently (We are working to obtain permission
from the government to use and display this data).
At this point, we heeded the advice of Heavner , who advised only making
attempts at C-I observations with a totally clear sky and distant (75-600 km)
thunderstorms, and made haste to the BSOS. The IMC-201 camera and related equipment
were set up, and directed towards roughly 240 degrees magnetic (actual measurements of
the azimuth were taken after the fact). A diligent watch was kept on the TV monitor to
see if a C-I event occurred. The system clock on the IMC-201 was not reset, but the error
was noted: 7 minutes, 32 seconds too fast when compared to GMT. All references to time
will henceforth be made based on system time to avoid confusion, but please keep the
actual time difference in mind.
c) Imagery / Information
This night was clearly the most prolific of them all. Only a few of these images
will be provided here: please see the following World Wide Web address for more
We will only show a few of the more interesting images here. More is available via the
internet at the above site.
The first event for the purposes of this summary occurred at 05:16:12.22 (Figures
10 and 11) according to the IMC clock. This was one of the more spectacular sprite
clusters of the night. It lasted for several video frames. Note the classic sprite structure in
figure 10 - well defined head and hair. The tendrils are not as clearly defined, which is
explained by the fact that we were quite some distance from the complex - 600 km away,
based on correlated radar, satellite, and azimuthal data. The altitude of the tallest sprite
was 85.5 km based on elevation derived from the video monitor (Please see Appendix B
for details on altitude calculations). Note the stars in the background of this image: the
sky was clear enough to see such distant sprites. Note the large lights to the extreme left
and right of the image, on the horizon. A persistent feature in all of these images, these
lights are simply housing complexes, both in Papillion, Nebraska, and in the Capehart
section of Bellevue, Nebraska.
Full Size: FIGURE 11
The next event occurred at 05:21:04.73. Visible here, in Figure 12, is a sprite
head, with hair, and a strange looking disk-shaped feature above the hair. This is clearly
the elve-type event described by Lyons and Dowden. Comparison with Lyons 
work tends to confirm this. The estimated altitude of the elve was 84.4 km, and the sprite
was 73.0 km (approximately).
At 05:22:36.00 (see Figure 13), a line or group of sprites was observed well above
the horizon. This was the best example of a line of sprites from the whole summer. For
comparison, note the radio beacon to the right on the image (just above the "36"). Here,
the estimated altitude of this line was 70.4 km.
Full Size: FIGURE 12
Full Size: FIGURE 13
After this time, a final, striking display came to finalize the observations for the
15th of August. At 05:25:12.53, another strange, amorphous, disc-shaped glow appeared
a few degrees above the horizon, as in figure 14. This is the second - and best - example
of an elve on this night. A mere instant later, an brilliant, seemingly explosive-looking
sprite (figure 15) followed the elve. The time interval is so short (less than 1/100th of a
second) that the video time counter had no opportunity to update, thus the same time is
displayed. The estimated altitude of the elve was 84.4 km. I made no attempt to compute
the sprite, because it is somewhat difficult to distinguish the bottom of the elve from the
top of the sprite.
Size: FIGURE 15
2. Events from August 22, 1996
a) Synoptic Weather Scenario
On 22 August 1996, 0000 GMT (Greenwich Mean Time), a low pressure system
was located over the eastern Dakotas, with a cold front extending through South Dakota,
western Nebraska and Colorado (see figure 16). A series of thunderstorms was
developing along the frontal boundary. Cells were observed from eastern Colorado, and
thence into Kansas and southwestern Nebraska (see figure 17) .
b) Location / Description of local conditions
becoming visible. Having reviewed the radar summaries and satellite imagery, we made
the decision to quickly set up the camera and equipment before the anvil came too far
east. After the fact, we obtained NLDN imagery courtesy of the Air Force Global
Weather Center (AFGWC).
measurements of the azimuth were taken after the fact). A somewhat less diligent watch
was kept on the TV monitor on this occasion: other efforts to record geographic
information was conducted while the video camera was playing. Also, a conventional
video camera was placed. alongside the IMC-201 in hopes of capturing a C-I event using
c) Imagery / Information
clearly identifiable. There are two other, smaller sprite heads on each side of the main
head. This sprite represents the typical sprite cluster observed during our observations.
Note the greater elevation at which this sprite was observed: the parent thunderstorm
complex was 412 km away - much closer than the August 15 event. The estimated
altitude for this event was 94.3 km.
phase. In addition to being diffuse, is not very bright, making it even harder to detect.
This demonstrates that there may be many more sprites than expected, since some, at
least, will be essentially sub-visual. The estimated altitude for this event is 105.6 km.
previously discussed C-I events. We discovered that we are in desperate need of some
type of cross validation check to provide a more reliable way of assessing the "identity"
of various atmospheric phenomena - many of the possible C-I events in question may
very well be simple cases of lightning cloud flash/reflection. The cases described above
(15 and 22 August 1996) were clear cut, given the clarity of the sky at the time of those
III. LESSONS LEARNED / AREAS FOR FUTURE IMPROVEMENT
A. Multispectral Data
imagery that could have been collected. Here, we will summarize the different types of
imagery that are recommended for further exploitation, in addition to brief explanation
regarding the utility of each.
1. Conventional visual data
have conventional visual imagery - such as from a normal video system- in order to
compare with imagery on the 7774 angstrom filter. In addition to assisting in the more
accurate identification of some events, if would have more readily screened out situations
like cloud flash, etc. Along with this, none of the events noted on either the 15th or 22nd
of August were seen with the naked eye. This was due, primarily, to the need to watch the
equipment of a near continuous basis.
2. Low-light Video
difficulties making connections work with existing components. We were able to re-
engineer the equipment successfully, but a better designed LLTV camera is needed to
optimize our ability to observe sprites and their structure. Also, the resolution and
focusing power of the video lens left much to be desired. One exciting possibility is the
utilization of modified astronomical equipment (telescopes, etc.). This kind of work is
being conducted already at the Lagmuir Laboratory in New Mexico. Exciting imagery has
resulted from this effort. We will need to find a way to record such telescopic imagery for
3. Multi-filter imagery
different spectral increments. Part of this process needs to be the utilization of the six
different filter wheels on the XYBION IMC-201 camera. Again, since we were operating
with only one camera in the field, we did not have the flexibility to look at different
wavelengths (with a rotating filter wheel) without sacrificing the quality and consistency
of having the 7774 angstrom data. In the future, we will need two LLTV cameras
operating in conjunction: one to record 7774 angstrom imagery, and the other to record
multi-filter imagery. This will certainly aid in future analysis.
4. Image analysis
correctly, thus forcing us to scrap that part of the project. We need to research a manner
in which to capture imagery in a digital fashion and perform image analysis. This function
will aid us in discriminating between real C-I events and cloud-flash (and other false C-I)
B. Utility of Digital Radar data and NLDN information
possible sprite producing thunderstorms. We were only able to save synoptic scale radar
summaries from the WSI Corporation. We need the ability to quickly obtain and store
near real time radar data for post-analysis.
Similarly, the NLDN products were useful, but only in a limited fashion. Like the
radar data, the NLDN was able to provide an advanced warning with regard to the
possibility of C-I genesis. It would be highly valuable to obtain and store this data at a
higher temporal and spatial resolution. The data we had access to from AFGWC was of
limited value: the imagery consisted of 3 hourly lightning summaries, leaving us unable to
correlate the C-I events we observed with actual lightning strokes. This is, I believe, a key
in understanding the interactive roles played by thunderstorms and the upper atmosphere.
C. VLF data correlation
Creighton, and future efforts will incorporate the VLF signal with the C-I imagery.
D. Additional resources needed
improve our C-I observation efforts.
of camera setup, then possibly more imagery would have been captured last summer. In
addition to this point, keep in mind that more people will be needed to handle multiple
cameras and other equipment, if these efforts are to expand in any of the ways
2. Power supply
would have power available for the equipment. Though it may seem a rather banal issue,
it is a significant limiting factor in planning where to go for potential observation sites.
Clearly, access to a portable generator of some type would aid in this effort, but short of
this, an active effort must be put forth to solicit any organization (or private citizen) to
allow access to an electric power supply. Bear in mind also that much of the equipment
(the IMC-201 camera, in particular) is a sensitive to power anomalies.
3. Location and Number of Observation Sites
observe C-I phenomena. It is not likely that most future observers will have an aircraft
available, so a careful search for site(s) on a high point, such as a hill / mountain is
needed. A given site is particularly desirable if it has a full view through 360 degrees of
azimuth, though most sites will not likely have such a helpful view. Therefore, we
developed a number of sites using the homes of private citizens (after gaining prior
permission) to observe as far as possible through 360 degrees. Any team trying to
conduct C-I observations needs to have this part of the problem researched before any
cameras are set up for recording.
E. Final thoughts on C-I Mechanisms.
Based on all the hypotheses presented, and the observational information
available, here is an overall hypothesis that we feel may explain sprites, in particular. We
have no distinct hypothesis on blue jets yet, and we need to obtain more information
regarding elves (thought the EMP theory seems plausible).
The entire consideration of sprites and blue jets requires us to take a step back.
Looking at figure 20, consider the overall charge distribution associated with the
The charges are first noticed in the ionosphere, where a variety of ions/electrons are
available in abundance. Next, the top of the thunderstorm possesses an overall positive
charge, while the lower portion of the cloud (around 6 km) has a strong NEGATIVE charge.
Below the 0 deg C isotherm, we will observe a small concentration of positive charge, and
the earth's surface will also possess a net positive charge.
argument is that the entire distribution of charges maintains an overall balance,
preventing a flow of charge in any direction. Of course, this is not a completely accurate
assumption: charge is being exchanged dynamically between the ionosphere and the
lower parts of the atmosphere. This assumption is only valid in the sense of time scale: in
other words, in the time span of a second or less, this statement is more or less acceptable.
Now, let us suppose that a discharge takes place between the negatively charged
region at 6 km to the ground. Consider, then, that the aforementioned balance of charge
has been violated, thus enhancing the electric field gradient between the top of the cloud
and the ionosphere relative to the lower part of this electric dipole complex. This allows
for a possible discharge to take place between the positively charged top of the cloud and
the ionosphere. It would be reasonable to argue that a discharge would take place in this
manner, since the electrical potential difference between cloud/ionosphere area would be
the same, while the potential difference between the cloud top and the lower part of the
cloud would be significantly decreased. This, then, is a possible scenario that we might
use to explain the appearance of sprites. Of course, the potential difference between the
cloud and will need to be high enough to cause dielectric breakdown of the intermediary
atmospheric gas - a distinctly feasible possibility. This hypothesis is inspired, in part,
from the work of Wilson [1925, 1956], though modified slightly.
observed the events at different times of the year, and under somewhat differing
conditions. Specifically, we observed obvious Red Sprites, (possible) blue jets, and
certainly at least two instances of elves. We were able to compute rough altitudes for the
tops of these flashes, and correlated them with distinct meteorological phenomena.
There is now a clearer sense of how to capture this imagery, and an improved plan
of attack for the systematic pursuit of these strange "lights-in-the-sky". We know what
conditions are best suited for sprite observation. We have a good idea of the kind of new
data and imagery that will be needed in the future to better understand these phenomena.
Finally, we have a working hypothesis to explore for future validation with
respect to the formation of these C-I events. As data collection efforts improve, we hope
to refine our understanding of these atmospheric events.