Normal-speed video

Prior to the sprite campaign conducted at Langmuir Laboratory in 1997, normal-speed video timing was obtained by placing one or more LEDs within the field of view and connecting the LED(s) to a calibrated GPS-based signal(s). The UTC time in hours, minutes, and seconds was either placed directly onto the video by a VCR or was added to the video in post-processing. The time placed on video was accurate to only a few tenths of a second. Sub-field accuracy (

17 ms) was obtained by estimating the time of the LED transition by implementing a procedure similar to that detailed in B.4.2.

A new technique was employed in 1997 to obtain normal-speed video timing. Horita GPT-50 units were used to place GPS information directly onto the video signals. The Horita units were specially modified by the manufacturer to insert a white vertical stripe onto the left side of the video image in response to a GPS-based pulse.

**Figure B.2:** Individual fields of a digitized normal-speed video sequence. The vertical stripe in the middle image was placed on the video signal at the moment of a second transition. The time shown at the top was the calibrated start time of light integration for the video image, while the time in the lower left is an approximate time updated by the GPT-50 every second. The light integration period actually preceded the second transition because images were not output to the video signal until after the light integration interval was over.
$\begin{figure}\begin{center} \par\epsfig{file=eps/horita_bar.eps,width=6in}\par\par\end{center}\end{figure}$

Figure B.2 shows 3 video fields extracted from a normal-speed video sequence around the time of a sprite event. The Horita GPT-50 inserted a vertical stripe at the left side of the middle image. The top of the stripe corresponds with the exact moment in which the onset of a second occurred according to GPS timing. By examining where in the video field the top of the stripe appears, one could theoretically determine the exact timing of the video to within the time period required to scan one line (

100 $\mu$ s). In practice, however, such precision has not been achieved since the output timing characteristics of the video camera were not known precisely.

An image field is not converted to a video signal until after the light integration period is over. Consequently, the vertical stripe will always appear in a frame prior to the one in which the transition actually takes place. The length of time it takes to output a video field as a video signal after the start of light-integration is always at least the duration of a video field;

= 16.7 ms. However, there are additional delays on the order of a few milliseconds. To determine the actual delay, light from return strokes within the camera's field of view was used. Because the light output from a lightning discharge increases many fold during the return stroke process, return strokes are easy to detect on video. The actual time of the return strokes could be determined precisely from the recorded sferics data. By examining many return strokes, the delay for the placement of the vertical stripe into a video image relative to the start time of light integration for the video image was determined to be 20 ms with an uncertainty of approximately 1 ms. Such precision was adequate for most normal-speed sprite observations.