On 25 April 1993 the Los Alamos National Laboratory launched a small
satellite, Alexis, into an 800 km circular orbit at 70
inclination.
Alexis carries a VHF radio receiver called Blackbeard that captures the
waveforms of electromagnetic radiation for later replay
[Massey and
Holden, 1995]. The major discovery by the VHF receiver was a phenomenon known
as Trans-Ionospheric Pulse Pairs (TIPPs)
[Holden et al., 1995], an example
of which is shown in Figure 1.
These pulses cover the full band width of Blackbeard from 28 to over 166 MHz,
last a very short time, about 4 
s, and are separated by about 50 s
[Massey and Holden, 1995]. The chirp
duration for the event in Figure 1
is somewhat longer than most, about
10 s. Since the lowest frequencies of TIPPs approaching the plasma
frequency
of the peak of the ionosphere exhibit the dispersion expected for a signal
passing through the ionosphere, the source of these signals
must be below the ionosphere. In its almost four years of operation, the
Blackbeard instrument detected over 1000 TIPPs.
Immediately it was conjectured that these events were associated with a lightning discharge process, and that the double chirp was simply the reflection of the initial pulse from the surface of the Earth [Holden et al., 1995]. However, a competing theory soon arose theorizing that the double pulse was created by upward propagating discharges in the middle atmosphere well above the clouds [Roussel-Dupre and Gurevich, 1996]. More recently, a double-pulse source within the tropospheric cloud itself has been suggested [Moldwin et al., 1996].
It is clear that the TIPP phenomenon is associated with cloud electrification. The TIPPs in the Alexis study period are predominantly over the main thunderstorm producing regions of the world and follow the seasonal variations in thunderstorm occurrence rate and location [Zuelsdorf et al., 1997]. However, the diurnal variation of TIPPs differs from that of cloud-to-ground lightning [Zuelsdorf et al., 1998a], and the rate of TIPP occurrence as detected by Blackbeard is much less than that of cloud-to-ground or cloud lightning [Zuelsdorf et al., 1998b]. With the express purpose of identifying specific lightning events with TIPP occurrence, we undertook a study of the records of the National Lightning Detection Network [Cummins et al., 1998] obtained for the time periods of overflights of the U.S. by the Alexis satellite. The statistical results of that study have been reported by Zuelsdorf et al. [1998b]. Since the NLDN stations studied were situated in the contiguous 48 states, we limited the TIPPs analyzed to the 17 that occurred while we had NLDN data (summer 1995) and when Alexis was over the central US.
In 10 ms a TIPP producing electromagnetic signal could propagate to Alexis from anywhere in its field of view, but within 10 ms of the 17 TIPPs there were only two simultaneous cloud-to-ground strokes, one each positive and negative. Since the NLDN stations miss few cloud-to-ground strokes of either polarity [Cummins et al., 1998], the lack of a clear association with negative cloud-to- ground strokes indicates that TIPPs are not the result of the most common form of lightning discharges. The lack of a clear association with positive cloud- to-ground strokes indicates that the Sprite-associated middle atmosphere discharge mechanism of Roussel-Dupre and Gurevich, [1996] is unlikely because Sprites have been associated with the more rare positive cloud-to-ground lightning [Boccippio et al., 1995]. The number of coincidences we found with the cloud-to-ground discharges would be expected to occur in our sample if the two phenomena were uncorrelated. We did however find a correlation with positive intra-cloud pulses. Seven of the 17 TIPP events were associated with a corresponding positive cloud pulse within the 10 ms prior to the TIPP detection. There is a less than 0.1% probability that these two phenomena are independent [Zuelsdorf et al., 1998b]. One of these seven TIPPs on July 3, 1995 at 0627 UT was accompanied by the detection of two nearly simultaneous (within 0.75 ms) positive pulses at neighboring NLDN stations, 490 km apart. This separation time is consistent with the propagation delay from a discharge located between the two stations. In this paper we examine the geographic location of that causative pulse, and its altitude.