DATA_SET_DESCRIPTION |
Data Set Overview
=================
This data set contains pre-encounter and encounter images taken
by the Stardust Navigation Camera during the encounter with
asteroid Annefrank. Images from the following sequences are
included in this data set (descriptions in this section were
provided by the NAVCAM Science Lead, Dr. Raymond L. Newburn,
Jr.):
2002-09-03: Image Sequence #32 (Images 301-324)
-----------------------------------------------
In preparation for engineering readiness tests utilizing the
asteroid #5535 Annefrank, as STARDUST once more approached the
Sun and Earth sufficiently to begin limited imaging, a first
test was made of the new pattern matching and windowing
software. Coincident with this, a series of geometric
calibrations was attempted, since those of June 2001 were not
totally successful. In addition a calibration lamp image and
four full frame fields were acquired at zero and thirty
degrees, one of the latter compressed. These were intended as a
modestly comprehensive check for contamination, for scattered
light, and for compression. It was found that there had been a
small amount of recontamination in the 10 months since the
previous image. This was most obvious in the calibration lamp
image. Star images remained sharp, with the same point spread
as earlier, but with a very shallow skirt of scattered light.
The pattern matching and windowing failed at 14 of the 19
angles. At larger scan mirror angles there was a problem with
increasing scattered light. The windows used were only 21x21
pixels, and it became clear that somewhat larger windows were
necessary and that there were still geometric calibration
problems. The contamination on the periscope was found to be
significantly reduced compared to that of two years earlier,
perhaps due to some evaporation of the condensate into the
vacuum of space.
2002-10-09: Image Sequence #33 (Images 325-345)
-----------------------------------------------
This series of images again was intended as a test of pattern
matching and windowing and to supply some geometric calibration
of the system. The camera was brought above freezing for 60
hours and then allowed to cool back to normal operating
temperatures in an effort to remove contamination before
initiating these exposures. The series consisted of 20 pattern
matching and windowing tests, ten each at 47.8 and 64.0 degree
scan mirror settings, each image consisting of four 41x41 pixel
windows, and one full frame image at a scan mirror setting of
15 degrees to check on the effect of a split field (partially
on and partially off the periscope). Half of the 47.8 degree
and all of the 64.0 degree tests were successful in locking up
on the desired pattern of stars, but the target stars still
were not well centered. These were engineering tests and led to
significant improvement in the software and to a better
understanding of spacecraft behavior and capabilities, but the
images provided little useful data for any sort of photometric
calibration following the fourth heating cycle that preceded
this series. The split image indicated that it should be
possible to use the periscope on Wild 2 approach as always
intended. Good geometric calibration of the periscope still
remains to be carried out, and the periscope was not used for
the Annefrank encounter. In the absence of any dust hazard, it
was not necessary to keep the spacecraft oriented along the
velocity vector, so Annefrank tracking utilized mirror angles
from 17.7 to 111.3 degrees.
2002-10-31: Image Sequence #34 (Images 346-350)
-----------------------------------------------
One of the goals of the engineering readiness tests on
Annefrank was to exercise the optical navigation team and to
attempt to improve flyby accuracy using optical data. The
approach to Annefrank was from a phase angle of 150 degrees,
unfortunately, which meant the asteroid would be poorly
illuminated and be very faint. There were no asteroid data
available for phase angles larger than 100 degrees. It was
assumed that Annefrank would be about 1.5 magnitudes fainter
than the nearly linear decrease of about 0.03 mag/deg that is
common to asteroids at smaller phase angles. It seemed that we
would have a fair chance of detection 38 hours before
encounter, which time was used for this first attempt. Five
images were obtained using three 151x151 pixel windows and
exposures of 1, 1, 2, 5, and 5 seconds. As we later found out,
the asteroid was much fainter than expected and the spacecraft
drift during the exposures (smear rate) much larger than we had
previously experienced. Further the camera pointing was not as
accurate as we had expected (the geometric calibration was not
yet solid). The use of new controller software, inadequate
settle times after attitude changes, and a larger moment of
inertia with the aerogel grid open were all suggested as
reasons for the drift and pointing problems. The cause is still
being investigated. This is why these tests were run, to make
sure something like this doesn't happen to us on Wild 2. The
bottom line is that the asteroid was not found in these five
E-38 hour images.
2002-10-31: Image Sequence #35 (Images 351-355)
-----------------------------------------------
A second set of approach images was acquired at E-32 hours. The
same windows and exposure times were used as for the previous
set at E-38 hours. The problems were much the same, as were the
results. Annefrank was not found.
2002-10-31: Image Sequence #36 (Images 356-360)
-----------------------------------------------
A third set of images was acquired at E-26 hours. The same
windows and exposure times were used as for the previous sets
taken at E-38 and E-32 hours. The problems were much the same,
as were the results. Annefrank was not found.
2002-11-01: Image Sequence #37 (Images 361-365)
-----------------------------------------------
Given the experience of the first three sets of approach
images, the navigators decided to increase the window size to
181x181 pixels and make all of the exposures 5 seconds for this
set at E-18 hours. The image smear can only be described as
horrendous. This doesn't matter for measurement purposes, IF
the target can be found. The asteroid was still several
magnitudes too faint for detection when smeared over some 20
pixels, and it was not located.
2002-11-01: Image Sequence #38 (Images 366-370)
-----------------------------------------------
A final set of approach images was attempted at E-12 hours.
This time all of the available communication bandwidth was
given to one window in one image (#368), making it 701x701
pixels. The other four images were given 3x3 pixel windows and
were retained only to avoid having to reprogram and transmit
too much last minute new command software. The smear was still
large (21.7 pixels) and the asteroid still was not located.
2002-11-01: Image Sequence #39 (Images 371-407)
-----------------------------------------------
Twenty-five minutes before the closest approach, images were
acquired to attempt autotracking. Pointing was based upon radio
navigation of the spacecraft and the best ephemeris for the
asteroid supplied by JPL's celestial mechanics specialists. By
this time the phase angle was down to 130 degrees and the range
was only 11,415 km. Annefrank appeared in the first image,
though far from centered. The navigators chose an exposure of
65 ms to make sure they were going ``deep enough,'' so the
images were well exposed. After the first few images, only
every third image was transmitted to the ground, the others
being used only to initiate autotrack. After 15 minutes, at a
range of 5434 km, exposure was reduced to 25 ms. In all, 15 of
37 images taken with 65 ms exposure were received on Earth. Of
these, the first two or three were partially on the periscope,
and three show a large amount of smear, but several are of
scientific use. Autotracking was initiated shortly before
reducing the exposure, and image 410 and all subsequent
Annefrank images are well centered in their frames.
2002-11-01: Image Sequence #40 (Images 410-445)
-----------------------------------------------
Exposure times on Annefrank were reduced to 25 ms beginning
with image 410 at a range of 5088 km and a phase angle of 113
degrees. Images beginning with #420 started to show saturation.
This was predicted, but these images were being taken to test
the autotracking rather than for scientific purposes, and
autotrack works perfectly well with saturated images. The
images soon reached 80% saturation, so images 420 through 445
are of limited scientific use. Every image was transmitted to
the ground beginning with #426, a total of 26 images with 25 ms
exposure. Twenty-two of these have some to nearly total
saturation.
2002-11-01: Image Sequence #41 (Images 446-476)
-----------------------------------------------
Beginning with image #446, exposure time was reduced to 5 ms.
In fact the characteristics of the shutter are such that
alternate images are given exposures shorter by 1.5 ms than the
set value, so in fact all even numbered images have an exposure
of 3.5 ms and odd numbered ones 5 ms. It was intended that
these images be of scientific as well as engineering use. If
Annefrank had not been acquired by this time, there was little
hope of acquiring it, so there was no need to saturate the
images. The subsequent images (through image 476) taken at
phase angles from 71.0 to 47.2 degrees constitute the best
images for scientific use. During this period the range fell
from 3133 km to 3078.5 km and increased back to 3162 km, so
there is minimal change to scale.
Processing
==========
The images in this data set were created by the DMAPKTDECOM
program developed by Applied Coherent Technology Corp, Herndon,
Virginia and operated by the Stardust Data Management and Archive
Team at JPL, Pasadena, California. This program assembled images
from raw telemetry packets sent down by the spacecraft and
populated the images labels with housekeeping values, decomutated
the binary image headers, and computed geometry parameters using
SPICE kernels. This program did not apply correction of any kind
to the image data.
In the cases when only certain sections of the detector were
downlinked, the program filled the pixels in the image
corresponding to the areas for which data had not been downlinked
with hex null values (i.e., zeroes). In such images window
objects define the areas containing non-null data.
Data
====
The images in this data set are in standard PDS format. Each file
includes an attached PDS label at the beginning of the file,
followed by a histogram, and ending with the image itself. The
PDS label contains two OBJECT definitions that describe the
storage requirements for both the histogram and image objects.
The label also describes the circumstances surrounding the
collection of the calibration image. This meta-data is in keyword
and value pairs and each of these keywords is described at the
end of this document.
Camera Description
------------------
The camera has a 1024x1024 array as the active portion of the
CCD. The images that are stored on this volume, however, contain
more than just the active portion of the CCD. Each line contains
a sync pattern, a line counter, 12 baseline stabilization pixels,
the 1024 pixels from the active portion of the CCD, and finally 8
over-clock pixels used to measure the quantum efficiency. The
number of rows for each image is always 1024, no matter what
compression mode is used, but the number of columns for each
image depends on the compression mode used.
Compression Modes
-----------------
The NAVCAM images can be either 8-bit or 12-bit data. The 12-bit
data is commonly referred to as 'uncompressed data', while the
8-bit is referred to as 'compressed data'. This compression is
accomplished by a 12-bit to 8-bit square-root look-up-table
compression method, which is implemented in the hardware of the
camera electronics. This compression is lossy and the estimate of
the 12-bit image can be recovered using the look-up table
mentioned in Appendix 3 of the Calibration Document. Both the
image and histogram portions of the data file require different
amounts of storage space, dependent on the compression mode used.
In uncompressed mode with 12-bit data, the pixels are expressed
in two bytes, as 16 bits per pixel. The upper nibble of the most
significant byte is always zero for these images. In compressed
mode with 8-bit data, the pixels are expressed in a single byte.
Number of Columns within Each Row
---------------------------------
The general form of each line for each image is fixed. The row of
data from the camera can be categorized into five different
regions:
1. Sync Pattern Always 2 bytes, with value 0x0000
2. Line Counter Always 2 bytes, values from 0 to
1023
3. 8 BLS pixels (*) Baseline Stabilization pixels,
either 1 or 2 bytes per pixel
4. 1024 image pixels (*) Either 1 or 2 bytes per pixel
5. 12 over-clock pixels (*) Used to measure quantum
efficiency, either 1 or 2 bytes
per pixel
(*) The pixels are either 1 or 2 bytes per pixel dependent on
the compression mode. Uncompressed, 12-bit images require
2 bytes per pixel, while compressed 8-bit images require 1
byte per pixel.
For the uncompressed, 12-bit data, each row contains 1046
'pixels' of data, which is exactly 2092 bytes. This is 2 bytes
for the sync, 2 bytes for the line counter, 8 pixels at 2 bytes
per pixel, 1024 pixels at 2 bytes per pixel and, finally, 12
pixels at 2 bytes per pixel. In equation form:
bytes_per_uncompressed_line = 2 + 2 + 2 * (8 + 1024 + 12) = 2092
For the compressed, 8-bit data, each row contains 1048 'pixels'
of data, which is exactly 1048 bytes. This is 2 bytes for the
sync, 2 bytes for the line counter, 8 pixels at 1 byte per
pixel, 1024 pixels at 1 bytes per pixel and, finally, 12 pixels
at 1 bytes per pixel. In equation form:
bytes_per_compressed_line = 2 + 2 + 1 * (8 + 1024 + 12) = 1048
Reading with RAW Image Readers
------------------------------
When using any of the supported PDS readers, this extra data at
the beginning and end of the line is not displayed, but when
reading these images with a raw raster-scan style reader, this
extra data at the beginning and end of each line must be taken
into account.
When reading images with raw readers, use the following values:
Compression Mode # Rows # Columns Data Type
---------------- ------ --------- -----------------------------
Compressed 1024 1048 BYTE data
Uncompressed 1024 1046 MSB_Unsigned_integer (16-bit)
Finding the Offset to the Data within the File
----------------------------------------------
When trying to read the histogram or image arrays from the file
using a RAW reader, the reader must first skip all of the
information before the object to be read. As an example, to
read the image object using a raw reader, the reader must first
skip the PDS attached header, as well as the histogram data. To
determine the amount of data to skip, examine two keyword pairs
from the attached label.
To advance to the beginning of the histogram data, examine the
following keywords:
RECORD_BYTES = 2092
^IMAGE_HISTOGRAM = 3
The first keyword defines the number of bytes within each
record, while the second keyword indicates at which record the
data begins. In this example, the data starts in record #3.
This indicates that 2 other records contain data prior to the
start of the histogram data. To compute the data offset,
account for 2 records of data: in this example, the offset is
(3-1)*2092 = 4184.
To advance to the beginning of the image data, examine the
following keywords:
RECORD_BYTES = 2092
^IMAGE = 11
As in the previous example, the first keyword defines the number
of bytes within each record. The second keyword indicates the
record at which the image data begin. To compute the data
offset, follow the example above:
Offset = ( ^image_histogram - 1 ) * record_bytes.
Example:
Offset = ( 11 - 1 ) * 2092 = 20920
Exposure Durations
------------------
The double-bladed shutter utilized by the camera has the property
that in one direction the exposures are 1.65 ms shorter than in
the other. Therefore a setting of 5 ms, which is the shortest
possible, results in alternate 5 and 3.35 ms exposures, those at
25 ms, alternate 25 and 23.35 ms exposures, and so on.
Occasionally bias frames, which do not require shutter action,
are transmitted to Earth. This changes the ``parity.''
While not always the case, for Annefrank image sequences the even
numbered images have the shorter exposures. The downlinked
exposures for these images have been adjusted and the correct
exposure duration values are provided in the labels.
Target Name in the Image Labels
-------------------------------
The target name in the image labels was set only for the images
where the target is either seen in the image or computed to be
with the camera field of view. For all other images the target
name was set to ``N/A''.
Consequently the label geometry items pertaining to the target
-- spacecraft-target position, velocity and distance, pixel
scales, and phase angle -- are only supplied for the images where
target name is not ``N/A'' and were computed for that specified
target.
Windowed Images
---------------
The IMAGE size parameter in the image label reflects the size
of the detector, however in some cases data from only certain
sections of the detector were downlinked. In these cases the
pixels in the image corresponding to the areas for which data
had not been downlinked are filled with hex null values (i.e.,
zeroes). WINDOW objects define the areas containing non-null
data.
Ancillary Data
==============
The geometry items included in the image labels were computed
using the following SPICE kernels archived in the Stardust SPICE
data set:
Kernel Type File Name
------------ ---------------------
LSK naif0007.tls
PCK pck00007.tpc
SCLK sdu_sclkscet_00074.tsc
FK sdu_v16.tf
IK sdu_navcam_v20.ti
SPK sdu_l_2002.bsp
CK (s/c) sdu_sc_rec_2002_v2.bc
CK (camera) sdu_nc_rec.bc
Coordinate System
=================
Geometric Parameter Reference Frame
-----------------------------------
Earth Mean Equator and Vernal Equinox of J2000 is the inertial
reference system used to specify observational geometry items
provided in the image labels. Geometric parameters are based on
best available SPICE data at time of image creation.
Epoch of Geometric Parameters
-----------------------------
All geometric parameters provided in the image labels were
computed at the epoch specified in the START_TIME label field.
Flip Required to Achieve ``As Seen By Observer'' Display
--------------------------------------------------------
Since the optical path of the camera includes a mirror and the
flight and image production s/w do not compensate for the flip
that this mirror introduces, the images displayed in normal
left-to-right sample, top-to-bottom line fashion have to be
transposed (flipped about left-top/right-bottom diagonal) in
order to appear as an observer located on the spacecraft would
see it.
Quaternion Provided in the Label
--------------------------------
The quaternion provided in the image label is an engineering,
or alternative, ``non-SPICE'' style, quaternion providing the
rotation from Earth Mean Equator and Vernal Equinox of J2000
inertial reference frame into the Stardust spacecraft frame.
Note that although this quaternion was downlinked in the binary
image header and included in the image label, it was not used
directly to compute any of the geometry parameters. Instead it
was converted to a SPICE-style quaternion and written to a
C-Kernel file, which was then used by the DMAPKTDECOM program
to calculate observation geometry.
Software
========
The images in this data set are in standard PDS format and,
therefore, can be viewed by a number of PDS-provided and
commercial programs. For this reason no special software is
provided with this data set.
Contact Information
===================
For any questions regarding the data format of the archive,
contact Stardust NAVCAM Science Lead:
Dr. Raymond L. Newburn, Jr.
Phone: +1 (818) 354-2319
Electronic mail address: Ray.L.Newburn@jpl.nasa.gov
MAIL STOP 264-379
Jet Propulsion Laboratory
California Institute of Technology
4800 Oak Grove Drive
Pasadena, CA, 91109-8099
USA
or Stardust Data Management and Archive Team (SDMA):
Charles H. Acton, Jr.
Phone: +1 (818) 354-3869
Electronic mail address: Charles.Acton@jpl.nasa.gov
Boris V. Semenov
Phone: +1 (818) 354-8136
Electronic mail address: Boris.Semenov@jpl.nasa.gov
MAIL STOP 301-125L
Jet Propulsion Laboratory
California Institute of Technology
4800 Oak Grove Drive
Pasadena, CA, 91109-8099
USA
|