Our identification of the location of at least one event provides strong support for the ground reflection model. The fact that it is a cloud pulse and not a cloud-to-ground stroke provides equally strong indication that the middle-atmosphere double-discharge model is not appropriate since upward-going discharges are associated with positive cloud-to-ground strokes [Boccippio et al., 1995]. Furthermore, the calculations upon which this initial hypothesis was based has recently been redone with significantly different results [Symbalisty et al., 1997]. For the tropospheric-cloud double- discharge mechanisms to be correct one would have to assume that the pulse separation was coincidentally the same as the echo delay expected for this event. It seems clear from this triangulated example and the untriangulated results of Zuelsdorf et al. [1998b] that TIPPs are generated in a single event near the tops of clouds followed by the reflection of that energy by the surface of the Earth. The inference by Zuelsdorf et al. [1998b] that the energy of the pulse does not travel much beyond the line sight from the cloud seems to be confirmed by our case study in which only two of the five neighboring NLDN stations detected the pulse that caused the TIPP under discussion.
If we examine the various electric phenomena that are associated with thunderstorms we can identify one that is a good candidate for being associated with TIPP production. Le Vine  found that the strongest producer of radio frequency noise had a bipolar structure in which a strong pulse was followed by an overshoot in the other direction over a period of 10-20 s. These pulses occurred in isolation. Later Willett et al.  studied these same pulses, naming them Narrow Positive Bipolar Pulses. Both conclude that these pulses are the strongest source of cloud generated HF radiation. However, Willett et al.  and Smith et al., [Direct isolated thunderstorm radio emissions, unpublished manuscript, 1998; hereafter referred to as SSHR98] agree that NPBPs have a weaker amplitude than cloud-to-ground flashes as measured when sampling the fast electric field. This is consistent with our limited detection by NLDN stations of TIPP correlated pulses, whereas cloud-to-ground pulses are detected at many stations. Furthermore, our correlated pulses do not exhibit a signal strength any stronger than the uncorrelated cloud pulses detected by NLDN stations [Zuelsdorf et al., 1998b]. That we are able to locate the TIPP source at an altitude of 8.0 0.7 km on an isochron in the vicinity of an active storm lends further support to the theory that TIPPs are associated with NPBPs [Smith et al., 1997]. In addition, the summation of our rise and peak-to-zero times is not inconsistent with the Willett et al.  measurement of the total duration including the undershoot of the pulse being 30 s or SSHR98 who measure 25.8 4.9 s. These observations appear to be entirely consistent with the narrow bipolar pulse source for TIPPs but, as in our study, they do not identify what process in clouds leads to the production of these very special, radio-frequency producing events.
In closing we note that we used the NLDN raw data not because it was ideally suited for the detection of the cloud pulses responsible for TIPPs. These data were chosen because of their retrospective availability and because we did not know with what cloud discharges TIPPs would correlate. The high bandwidth recordings of waveforms such as those performed most recently by Smith et al.  are much more suited to the study of these pulses. The short duration of these bipolar pulses suggest that the NLDN stations that do not record pulses of duration less than 4 s could miss many pulses especially close to their source before they dispersed. Moreover, on strong short pulses it is possible that NLDN stations could trigger on the long negative "undershoot" portion of the bipolar pulse. We also note that the signal may propagate over the horizon from the cloud pulse for stronger events. The radio horizon is further than the line of sight optical horizon and a "ground wave" may also extend the range of the signal [Proctor, 1995]. Nevertheless, the coincidence of cloud to ground activity with our isochron deduced solely from the two cloud-pulse delays gives us much confidence that we have indeed found the geographic location of the source of at least one TIPP detected by Blackbeard.