Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/ops/stories/15.text Дата изменения: Wed Jul 14 15:24:06 1999 Дата индексирования: Sun Apr 10 14:27:38 2016 Кодировка: Поисковые слова: storm

Discrete Gemoetry Moving Target Processing

There once was a mathematician who had to ride a lot of taxxis in a
big city. On one of these taxi rides, he noticed that the taxis
seemed to prefer to pick up or let off passengers only at
intersections. He wondered what a universe consisting only of the
intersections would look like. He then developed, strictly for
amusement, an entire non-Euclidian geometry based on such a universe.
This was called Taxicab Geometry. Later this formalism became the
basis of digital signal processing. Quantization, samples, pixels -
points with nothing in between.

Well, now, if you generalize this geometry by assigning a vector to
each point, you have the method we originally used to observe moving
targets.

We had the MOSS system, flight-tested before HST was launched during
the Voyager-Neptune encounter, which could calculate the exact
position and motion of anything in the solar system. But the rest of
the ground system was not sufficiently developed to handle moving
targets. It could, however, handle a spatial scan. So we rewrote
each moving target proposal to observe a fixed target with spatial
scan calculated to mimic the motion of the planet.

This required at least two iterations. The first one had a point and
scan which was "roughly" correct, based on where we thought the
observation would be scheduled. We (the old Science Planning Branch)
then delivered the proposal to SPSS, who scheduled the program on an
SMS. They sent us back the details of the schedule, and we then sent
back updated coordinates (RA and Dec), spatial scan parameters, and
the distance to the target in units of inverse meters. (Inverse
meters!!) I wrote a little FORTRAN program that would take in MOSS'
output and calculate these values. The second iteration required very
short turnaround.

With this technique, we did the observations of the great Saturn
Storm, which brewed up suddenly, and by getting everybody in SPB,
SPSS, and other branches to drop everything, we completely processed a
new proposal to observe the storm in about two days flat. That was an
astonishing accomplishment for the time when things normally took
weeks. I've often wondered if one of the seeds of the concept of
end-to-end processing that became Presto wasn't planted by that
excercise.

Nowadays, to process a moving target, one accomplishes all this by
simply pressing the "MOSS" button. But under the hood, the system
still computes fixed targets and spatial scans in order to track a
moving target.

Taxi!

Sure, where ya goin?

Jupiter.

!!!