DATA_SET_DESCRIPTION |
The data producer originally supplied a data set description.
However, a lien from the peer review requested a more detailed
description. PDS-SBN enhanced the description with information
provided in an unpublished paper about these data by
Woodney et al. (1998) [WOODNEYETAL1998].
Data Set Overview
=================
This data set contains digitized photometry of the reflections off
Io and Europa of the impact fireballs produced as the D, E, K, and N
fragments of comet D/Shoemaker-Levy 9 hit the atmosphere of Jupiter.
The data were acquired during the predicted impact period for each
fragment, from 17 through 20 July 1994.
Observations
------------
The following paragraphs were paraphrased from Woodney et al. (1998)
[WOODNEYETAL1998].
At Mt. Singleton in West Australia, the University of Maryland/Perth
Observatory team used a high-speed occultation timing photometer and
a 14-inch Celestron to look for flashes off Io or Europa during the
expected impact times of the D, E, K, N and P2 fragments. During the
D, E, N, and P2 impacts, Io was observed, while Europa was the target
for the K impact. This fragment was predicted to illuminate Europa
while the moon was in eclipse with Jupiter, possibly making the
satellite temporarily visible from Earth. The flashes from Io
were expected to be small compared to the reflected solar component
(one to ten percent), so the best opportunity to observe an effect
was during an eclipse.
The photometer had a bi-alkali photocathode, and the observations were
made with no filter. The resultant bandpass of the whole system
ranged from 312 to 615 nanometers, with an effective wavelength of
453 nanometers. The team obtained continuous light curves of Io at
the predicted times of the impacts of fragments D and E. Mechanical
problems during the impacts of K, N and P2 left large gaps in the
temporal coverage. This severely limits the usefulness of these data.
In particular, the P2 data were not included in this data set because
this coverage was very sparse. Therefore, only the following impact
data are included in this data set:
Start Stop
Impact Date UTC UTC Target Observing Notes
------ ------ ----- ----- --------------- ------------------------------
D Jul 17 11:30 12:00 Io Photometric sky
E Jul 17 14:45 15:15 Io Photometric sky, some wind
K Jul 19 10:07 11:10 Eclipsed Europa Mechanical problems, data gaps
N Jul 20 10:10 10:21 Io Mechanical problems, data gaps
A set of calibration observations was recorded around impact D.
These data are also included in this data set.
For each observation the appropriate moon or calibration source was
centered in a 30-arcsecond aperture. The observer tracked the moon by
looking through a 5-inch finder scope and using a hand paddle to drive
the telescope. This method, combined with the Jovian scattered light,
is one of the larger sources of error in the photometer data. Motion
of the moon within the aperture caused variations on the order of five
percent as more or less scattered light entered the aperture.
For each observing run, a voltage-to-frequency converter was used to
record the following data onto a four-track audiotape:
1. Data counts from the target, recorded as a frequency
2. WWV time signals (http://tf.nist.gov/timefreq/index.html)
3. A 60Hz reference signal
4. Voice comments of the observers
To optimize the system for detecting small events, the gain for the
voltage-to-frequency converter as high as possible, which caused the
sunlit satellites to be near the saturation point. When the converter
saturated, the frequency went to zero, which led to low values for
all saturations when the data were digitized.
During data collection, the observers recorded the UTC time associated
with selected second ticks from the WWV and the 60Hz reference signal
so that data could be synchronized.
Results from Impact D
---------------------
The following paragraphs are paraphrased from Woodney et al. (1998)
[WOODNEYETAL1998].
During impact D, the team observed a sunlit Io. The team had hoped to
hoped to observe the flash as a percentage increase in the brightness
of the moon. Data acquired near the time of the D impact are the best
set obtained by the team. The sky was photometric, tracking was good,
and there were no mechanical failures. No flashes were large enough
to be seen on a strip chart produced at the time of the observation.
However, the digitized data were carefully inspected for the several
minutes around the time of the first infrared precursor flash reported
by the observers at the Anglo-Australian Telescope (AAT).
At 11:54:20 UT, 26 seconds before the infrared precursor flash was
observed at the AAT, a small rise is seen in the digitized data. The
signal rises to about 7 percent above the average over three seconds,
plateaus for almost two seconds, then quickly drops back. While this
event seems too long to be noise, the team could not conclusively
determine if any reflection off Io of the D impact flash was observed.
Results from Impact E
---------------------
The following paragraphs are paraphrased from Woodney et al. (1998)
[WOODNEYETAL1998].
During impact E, the team observed a sunlit Io. Again, observing
conditions were good but the photometer data included many saturations
which made it difficult to analyze for evidence of flashes. However,
At the IAU meeting in August 1994, N. Raghavan, of the Nehru
Planetarium in New Delhi, reported photographing a flash off Io which
started at 15:14:42 UT and lasted three seconds. The team reanalyzed
the digitized, high-speed photometer data and found saturations from
15:14:40 to 15:14:46. Since the saturations occur at the same time as
the suggested flash, it is possible that the flash caused the last few
seconds of saturated data. Assuming the flash was as the saturation
level, the team estimated it to be 15 percent of the brightness of Io.
It is important to note that the start and stop times given in the
time-sychronized table for impact E are 11 minutes sooner than those
reported by Woodney et al. 1998 [WOODNEYETAL1998]. PDS-SBN could not
reconcile either set of times with the UTC of the possible, saturated
flash from impact E reported above. The data table for impact E should
be used with caution.
Results from Impact K
---------------------
The following paragraphs are paraphrased from Woodney et al. (1998)
[WOODNEYETAL1998].
The team proposed impact K would provide the best chance to observe
a flash because the target, Europa, was in eclipse with Jupiter during
the event. There was not reflected sunlight to overwhelm a lower
intensity flash. While data for this event are similar in quality
to the data for E, there are many saturations caused by scattered
light from Jupiter entering the aperture. The team was not able to
find evidence of a flash in these data.
Results from Impact N
---------------------
During impact N, the team observed a sunlit Io but high winds and
equipment failures resulted in sparse, poor quality data. However,
these data could be digitized and are included in this data set.
Processing
==========
For each observing run, the following data were recorded directly
onto a four-track audiotape:
1. Raw data counts from the target as produced the photometer, then
recorded as a frequency
2. WWV time signals, including seconds ticks
3. A 60Hz reference signal
4. Voice comments of the observers
The audiotapes were digitized in the fall of 1994 at Lowell Observatory.
Each four-track audiotape was digitized, at a rate of approximately 100
samples per second. The outputs from this process are ASCII tables which
contain only two columns:
1) Column 1 contains time stamps produced by the digitization process.
It contains a series of ones and zeros. Each occurrence of a one
represents the beginning of one second of time, corresponding to
a specific clock time as recorded on the audiotape. The WWV time
signals were used when available. For the K fragment, the WWV was
not strong enough on the audiotape for the voice track was used.
For samples between the seconds ticks, the digitization process
output a zero in this column. There are approximately 99 samples
between each second. For the first record, the digitization process
output a value other than zero or one in this column. For the
time-synchronized version of a table (see the explanation below),
the process output a number greater than 63000 in this column to
mark the start of the time-synchronized samples.
2) Column 2 provides the intensity of the digitized raw data numbers.
Tapes were generally digitized twice: Once with the time reference
beginning at an exact UT time (this is the time-synchronized version
as noted in the label) and once with the time reference beginning at
the start of the data. By comparing the two, the exact timing of the
entire data set can be determined. If there was not a clear WWV signal
for a data set, or exact timing was not important for a set of
observations such as calibrations, there may not be a time-synched
digitization for that data set.
The data provided in this data set are raw and have not been processed
beyond the digitization procedure described above. Woodney et al. (1998)
[WOODNEYETAL1998] provide a recipe for converting the observed flux
increases into flash luminosities.
Parameters
==========
The following naming convention is used for the ASCII data tables that
contain data from impact D, E, K, or N:
TABLfns.TAB where f = D for impact D data
E for impact E data
K for impact K data
N for impact N data
n = 1 or 2 for the first or second half of
a set of observations
s = T for the time-synchronized version of
the impact data; if T is not present,
data file is not time-synchronized
Two calibration tables were provided: TABLD1.TAB and TABL2D1.TAB.
While these tables contain observations of standard stars, the data
provider did not list the targets. These calibration tables are not
time-synchronized. TABL2D1.TAB is simply a second digitization of
TABLD1.TAB that appears to be noise.
In addition to the impact and calibration tables, three tables of
unspecified targets are provided: TABL1.TAB, TABL11.TAB, and TABL12.TAB.
These data were used to test the digitization of the data using
different gains. TABL12.TAB has the correct gain.
|
CONFIDENCE_LEVEL_NOTE |
Confidence Level Overview
=========================
These data were peer-reviewed in 2002. During the review it was
noted these data are unique and should be archived. Reviewers
indicated these completely raw data could be reduced if someone
was determined to do so.
Data Coverage and Quality
=========================
Data from impacts D and E data cover the times of the impact events and
are reliable. However, mechanical problems during K and N impacts left
large gaps in the time coverage that severely limits the usefulness of
these data.
For each observation the moon was centered in a thirty-arcsecond
aperture. An observer looking through a five-inch finder scope
manually tracked the target by using a hand paddle. This tracking
method, combined with the Jovian scattered light is one of the larger
sources of error in the photometer data. Motion of the moon within
the aperture caused variations on the order of 5 percent as more
or less scattered light entered the aperture. The moon was also
occasionally lost from the aperture for brief periods. Additionally
there was no protection from the wind for the telescope and this
caused some interference with tracking.
Specifically, the numerous saturations seen in the data for impact E
are attributed to wind buffeting the telescope during these observations.
The wind made it difficult to keep Io centered in the aperture,
resulting in strong contamination by scattered light from Jupiter.
Since the detector was working very close to its limit, this additional
light was enough to saturate the detector.
Also, for impact E, the start and stop times given in the time-
sychronized table included in this data set are 11 minutes sooner than
those reported by Woodney et al. 1998 [WOODNEYETAL1998]. PDS-SBN could
not reconcile either set of times with the UTC of the possible, saturated
flash from impact E reported by Woodney et al. 1998 [WOODNEYETAL1998].
Therefore, the data table for impact E should be used with caution.
Finally, an unsynchronized and a time-synchronized data table were
provided for impacts D, K, and N. For impact E, only a time-synchronized
data table was provided.
Limitations
===========
One must be careful when interpreting these data. Most, if not all
changes in these light curves may be due to tracking problems.
To optimize the system for detecting small events, the gain for the
voltage-to-frequency converter was set as high as possible, which
caused the sunlit satellites to be near the saturation point. When the
converter saturated, the frequency went to zero, which led to low values
for all saturations when the data were digitized. Therefore, this is an
ambiguity between low values due to saturations and low values due to
the loss of the satellite from the aperture. Values that are below
the sky level are clearly saturated. However, some portion of all
low spikes that are still above the sky can be attributed to either
saturation or loss of a portion of the moon from the aperture.
The data provided in this data set are raw. While Woodney et al. (1998)
[WOODNEYETAL1998] provide a recipe for converting the observed flux
increases into flash luminosities, there is not clear description
of the steps required to calibrate the raw data.
|