ASCA Attitude Leap second problem
Last updated 2005-08-22
Throughout the ASCA mission life, the leap second was not taken into account
in the attitude determination, whereas there were five osscasions of leap second
insertion from 1993 to 1999. Consequently, near the end of the mission,
the ASCA attiutde was wrong by 5 seconds (i.e., the attitude in the attitude file
was that of the 5 second ahead).
All the attitude files in the rev_2
processing are affected by this bug.
All the pointing and slew attitude files are recreated fixing the leap-second bug.
New files are found at
ftp://legacy.gsfc.nasa.gov/asca/data/attitude_2005-April.
For pointed observations, the difference is completely
negiligible. On the other hand, the attiuded during the slew/maneuver operation was
significantly affected. In fact, this problem was realized when the ASCA slew data
was being analyzed. After the leap second correction, ASCA attiude during the slew
is correct within 0.1 deg, which is confirmed by source positions detected
druing the slew. (Published in proceedings
for the "MAXI workshop on AGN Variability" meeting in March 10 to 11, 2001, in Nikko, Japan).
Detailed explanation of the problem is found
in the Japanese report by Ryuichi Fujimoto on 2001-12-3.
In the following diagram, the data flow of the ASCA attitude determination is illustrated.
The attitude determination is made by proprietary software by the NEC contractor, which adopts
UTC, not ASCATIME. There was a bug in step 3 ALWAYS, when converting UTC back to ASCATIME.
Other minor bugs were also present in steps 1, 2 and 4 in some limted periods.
Before 1993/06/30 23:59:60
No effects of leap seconds. Attitude files are correct.
Leap second at 1993/06/30 23:59:60 to 07/01 00:00:00
Bugs 3 and 4 affect the attitude files, but they cancel for the keywords DATE0,TIME0, DATE1 and TIME1.
The leap second took place in the attitude file fa930630_1425.0231.
Afte the leap second insertion (ASCATIME = 15638403), 1 second has to added to the TIME column
value and MTOME1 value in the attitude file.
From now on, a one second has to added to TIME, MTIME0 and MTIME1 values of the attitude
files until the next leap second insetion.
Leap second at 1994/06/30 23:59:60 to 7/01 00:00:00
The leap second took place in the attitude file fa940630_1323.1620.
Afte the leap second insertion (ASCATIME = 47174402.5), 2 seconds have to added to the TIME column
value and MTOME1 value in the attitude file.
From now on, two seconds have to added to TIME, MTIME0 and MTIME1 values of the attitude
Leap second at 1995/12/31 23:59:60 to 1996/01/01 00:00:00
The leap second took place in the attitude file fa951230_1625.1222.
Afte the leap second insertion (ASCATIME = 94649498), 3 seconds have to added to the TIME column
value and MTOME1 value in the attitude file.
From now on, three seconds have to added to TIME, MTIME0 and MTIME1 values of the attitude
Leap second at 1997/06/30 23:59:60 to 07/01 00:00:00
There was a delay to update the leap second table in the ISAS database,
such that between 1997-07-01 to 07-09, the leap second was not taken into
account in the step 1 and 2 in the diagram above (see Fujimoto report for detail).
The leap second took place during the period of fa970629_1345.0030.
For this file, three seconds should be added to TIME, MTIME0 and MTIME1, and
DATA0/TIME0, DATE1/TIME1 should be modified accordingly.
The telemetry files ft9702_0340.0346m and ft9702_0346.0950
were created during the problematic period 1997-07-01 to 07-09.
For these files, one second has to be added to TIME, MTIME0, MTIME1,
DATE0/TIME0 and DATE1/TIME1.
For the attitude files after fa970629_1345.0030, four seconds have to
be added to TIME, MTIME0 and MTIME1 values.
Leap second at 1998/12/31 23:59:60 to 1999/01/01 00:00:00
The leap second took place in the attitude file fa981231_0000.2332.
Afte the leap second insertion (ASCATIME = 189302410), 5 seconds have to added to the TIME column
value and MTOME1 value in the attitude file.
From now on, five seconds have to added to TIME, MTIME0 and MTIME1 values of the attitude
After 1991 November
Y2K fix was made so that the date format was changed from dd/mm/yy to yyyy-mm-dd.
Correspondingly, the bug 4 in the above diagram was fixed.
Consequently, bugs 3 and 4 no long cancel, and the five seconds have to added not only to MTIME0 and
MTIME1 but also DATE0/TIME0 and DATE1/TIME1.
Script to fix the problem
For all the attitude files,
the attitude leap second problem was fixed with correct_leapsec.pl.
This is a perl script to call Perl CFITSO and ftools (sec2time etc).DATE0/TIME0 and DATE1/TIME1
(in UTC) are calculated from MTIME0 and MTIME1 (in ascatime) using the "sec2time" ftool which
takes into account the leap seconds (and YYYY-MM-DD notation was adopted).
Attitude redetermination of fa981108_0637.2040
Only for the fa981108_0637.2040 file, the attitude was redetermined at ISAS (but the leapsecond problem was
not fixed). The new file was pesonally sent from ISAS to Koji Mukai.
We have corrected the leap second problem on this file, and put at
ftp://legacy.gsfc.nasa.gov/asca/data/attitude_2005-April/POINTING/fa981108_0637.2040.gz.
Original attitude file sent from ISAS to Koji (i.e., attitude redetermined, but leap second problem not fixe) is kept as
ftp://legacy.gsfc.nasa.gov/asca/data/attitude_2005-April/POINTING/fa981108_0637.2040.original.gz.