February 25th, 2011

The last nine days have been a fairly boring routine as I grind along, running the data. There are two computers at work. The main computer (the one on which I’m writing this) analyzes the individual frames from the videos to identify Leonids. Thus, its input per run is about 54,000 individual frames of video; its output is an XML file containing data on the Leonids discovered in that video. It also outputs the same information into a plain TXT file, which can be read directly into a spreadsheet. I insisted on the use of the XML format because it is more robust than the TXT file; if someday in the far future, somebody comes across the TXT file, they’ll have no way of knowing what all the numbers mean, but if they come across the XML file, they should be able to figure out what the numbers mean. Here’s a modified representation of a single Leonid in the TXT file:

AL50R99111801570311 8 7425 true true 0.67
461.7 445.0 7.6 172.1 112.9 579 37.10 27.48 1773 10 0.42 false
463.7 448.6 7.4 172.0 111.2 580 37.11 27.50 3444 19 0.40 false
466.0 452.2 7.2 171.8 109.4 581 37.13 27.51 3152 16 0.19 false
468.1 454.9 7.1 171.7 107.7 580 37.14 27.50 4014 22 0.61 false
470.6 459.5 6.8 171.5 106.0 584 37.16 27.55 3253 17 0.64 false
472.9 462.5 6.6 171.4 104.2 584 37.17 27.55 2292 13 0.53 false
474.8 466.2 6.4 171.3 102.5 586 37.19 27.57 1909 14 0.65 false
476.2 468.8 6.3 171.2 100.7 585 37.19 27.56 2031 17 0.42 false

The first line presents first the identifier for the Leonid. “AL50R” means that this camera was aboard the Aria research aircraft [“A”], looking out the left side of the aircraft [“L”], pointed upwards 50º [“50”], and facing rearward [“R”]. The time of the first flare was November 18, 1999, at 01:57:03:11 Universal Time. (The final “11” is the frame number, pinning this thing down to 1/30th of a second). This Leonid was comprised of 8 flares, which together had an aggregate luminance of 7425. It began off-screen and ended off-screen (a very rare condition -- this Leonid must have been right along the edge of the frame). The accumulated positional errors for this Leonid added up to 0.67 -- that’s pretty good.

The next 8 lines present information on each of the 8 flares that constitute this Leonid. The first flare was at frame coordinates X = 461.7, Y = 445.0. This represents an angular altitude of only 7.6º and an azimuth of 172.1º. Calculations put it at a linear altitude of 112.9 km and a range of 579 km. The latitude of this flare (more precisely, the spot on the ground directly underneath it) is 37.10º North, while the longitude is 27.48º East. Its total luminance is 1773 and it was composed of 10 pixels. Its net error was 0.42, and it was NOT “byed” in (Byeing is my process for including a flare in the Leonid if it happens to pass directly over a star, which creates an ambiguous situation: did the Leonid disappear before going over the star, or is it merely lost in the star’s image? Byeing assumes that it’s still there, but I record that fact so that users of the data can adjust their analysis if they don’t want byed flares to be included.)

It turns out that some of the XML files are so big that my program can’t handle them, so in some cases I had to break datasets up into smaller datasets.

I have begun the preliminary work on analysis. It’s not much: just putting the datasets from a single camera together into a single large dataset. I then run the first-round statistical analysis on that data. I carry out a considerable amount of first-round analysis; it is not meant to provide final answers, but rather to provide us with the basic information we need for detailed analysis. So far I have found a number of interesting trends, but I’ll not publish them here until I have more confidence in them.