User's Guide to the DEBRIS Package
1. Introduction
This guide provides a description of the procedures involved in detecting
geostationary/geosynchronous orbital debris using an untracked observation
method. It is a general outline of how data is acquired and how it is
subsequently analyzed using the tasks of the DEBRIS package. A detailed
description of the individual tasks can be found in the help documentation
of the package.
2. The Method
In order to observe geostationary objects there are two obvious methods:
- Tracking
- If the telescope drives are on during the observations then of course all
geostationary objects will appear as straight lines in the east-west
direction. This means that the already faint objects get blurred out over many
pixels, making them even harder to detect.
- Untracked
- However, if the telescope drives are switched off during the observations then
all truly geostationary objects would appear as point sources. Unfortunately,
orbital debris cannot be assumed to be truly geostationary due to changes in
the orbit's inclination and ellipticity. Therefore it must be assumed to change
both its HA and DEC. Although this motion is comparatively small it is obvious
that this poses a limit on the exposure time if one keeps in mind that the
whole purpose of this observing method is to keep the object if possible in one
pixel.
The DEBRIS package was designed as a facility to analyze data acquired
using the untracked method. The idea is to compensate for the above mentioned
limit on the exposure time by taking a whole stack of frames and combining them
later thus making objects visible that were undetectable in a single image.
The following will identify the two major problems with this idea:
- - Since the observer has no idea how fast and in what direction a particular
piece of debris is moving it is impossible to observe this piece in the same
pixel in every image of the series.
- - Even if the first problem was somehow solved how can a detection program
(we are interested in an automated search algorithm!) distinguish between a
star and orbital debris ? (The roundness parameters of the DAOFIND task alone
are not enough.)
The first problem is unsolvable. One has to live with it. This means that
the telescope stays fixed with all drives switched off for the entire series of
observations. Before combining the images then they will be shifted with
respect to each other in many different trial directions and at many different
trial "speeds" in an attempt to simulate the motion of a particular object and
to bring all its light into a single pixel. At the end of this one is left
with very many combined images to be searched. A particular object will only
appear in very few (one or two) of these images, namely the ones which simulate
its motion best.
The second problem can be dealt with in a more elegant manor simply by
using an appropriate rejection algorithm when combining the images. In this way
one can achieve that the stars disappear almost altogether. So the problem of
how to distinguish between stars and orbital debris does not pose itself in the
first place.
3. The Data
As described above, the input to the DEBRIS package consists basically of
a stack of images that were taken with all telescope drives switched off.
The best results are expected from observations close to 0 degrees declination.
It is important to note that all images should have the same exposure time and
must be equally spaced in time. All image reduction should be performed
before entering an analysis, i.e. the images should be biassubtracted,
corrected for dark current, flat fielded and if necessary illumination
corrected.
(See demonstration image demo1.)
At present, the images should also have a particular orientation:
north <--> left, east <--> up, south <--> right, west <--> down.
The images do not have to be oriented in this way but it makes life a lot
easier if they are.
4. The Analysis
The first step is the shifting and combining of the images. This is
performed by the task DEBMV. The shifting for a particular image is performed
according to three parameters which represent a trial motion. The output
consists of the combined images
(see image demo4)
and various logfiles. The logfiles will be
used at later stages of the analysis and should not be deleted or tampered
with light heartedly. The same goes for file names: The DEBRIS package uses
easy to understand naming conventions for image-, log- and coordinate-files.
In this way a task will automatically find the files it needs without having
to bother the user. DO NOT rename files unless you know exactly what you are
doing.
The second step is the detection of orbital debris. The task DEBFIND scans
the output images of DEBMV for local intensity maxima using the task DAOFIND
in the APPHOTX package. Remember that stars have mostly been removed from the
image yet residuals may still lead to unwanted detections. In many cases
however these can be eliminated by using the roundness parameters of DAOFIND.
The output of this step consists of one
coordinate file per input image. Most
of these will be empty (except for the header). Since there are a great number
of coordinate files it would be a tedious task to check these files for
detections one by one.
The third step is therefore the reporting of the
results by DEBREPORT.
This task reports in which images objects have been found, the magnitude of
the objects and in how many "neighbours" a particular object was found. A
"neighbour" of an image is an image with neighbouring shifting parameters. This
quickly shows the user whether detections in two different images refer to the
same object or not. It is also some sort of measure of the reliability of a
detection.
4.1 DEBMV and DEBFIND vs DEBMVF
The task DEBMVF is a combination of the tasks DEBMV and DEBFIND. It serves
exactly the same purpose but there is an important difference in the order in
which the steps are executed. If DEBMV and DEBFIND are used for an analysis
then first all images are created and then they are searched. This is
impractical since it requires a large amount of available memory. Instead,
DEBMVF searches every image right after its creation and is then free to delete
the image if the user so wishes. So the actual analysis is best performed
using DEBMVF deleting every image after it has been searched. Any image can
be created again at any time using the information in the
logfiles; if one
wishes to create an image again just for the purpose of examination DEBMV
should be used. DEBFIND is best used for searching images repeatedly with
different parameter settings.
Throughout all documentation on the DEBRIS package almost all references
to the DEBMV and DEBFIND tasks apply equally to the DEBMVF task.
5. Summary
The "course of action": The stack of input images are shifted with respect
to each other in order to simulate the movement of a supposed piece of orbital
debris. Then the images are combined using a rejection algorithm so that stars
disappear. The combined images are searched for remaining brightness maxima and
the result is reported in a manageable fashion.
Detailed descriptions of the individual procedures can be found in the
respective help pages. Additionally, the help pages of the tasks IMCOMBINE
and DAOFIND provide detailed information on the rejection and detection
algorithms employed.
Back to The DEBRIS Page.
Joe Liske
Imprint