JAC Style guide: template
 Derivation of Statistical Information
This note provides some explanation on the method of production of the
weather and fault statistical summary for the performance of the JCMT. There is very little manual intervention in the generation of these tables, other than to correct obvious errors, mis-labelled categories, or to complete missing entries (where they can be found from other sources).
o The data are extracted from the reports completed by the telescope
operators at the end of their shift, one report for each shift (evening or morning); the shifts are normally of 8 hours duration. Because the change-over of TOs does not occur on the shift boundaries, the shift information is handed over to the following operator who will file the report at the end of shift.
o A completely separate fault reporting system is used by the TOs and other staff to record time lost to faults (including problems which have zero-time lost). This system is used by the staff to identify, trouble-shoot and solve the faults. Each fault should then have an appropriate solve report attached for future reference. This system is not used for generating statistics.
o Input from the TO reports comprises
- - the actual time scheduled (normally 8 hours);
- - any extension of this (normally due to good weather
and lack of pressing daywork);
- - loss of time on primary and backup programs due to the weather.;
(NB : in flexible observing mode, the move from a high-frequency
primary program to a low-frequency primary program because
of a deterioration of the weather conditions does not result
in any entry in the 'loss to the primary program' category.
- - loss of time to primary and backup programs due to faults,
divided into 6 categories : ANTenna, CARousel, INStruments,
COMputer, SOFtware, and OTHer. These categorizations are performed
by the TO at the time of filing and persist in the analysis,
although suggested changes in categories are suggested by the notes
to the analysis.
(NB : faults are defined as being in respect of subsystems
that have been commissioned are therefore expected to work
flawlessly. If the instrument under commissioning has a fault,
this is not recorded in the log.
(NB : previous correlations between faults as reported in these
TO reports and via the separate fault reporting system show high
levels of completeness. A similar correlation is also found between the TO reports and any completed Observer Reports for the period.
- - automated retrieval of weather conditions for the shift in question.
These are not analysed further and are not further correlated with
the reported conditions.
o Electronic submission of each report automatically triggers the
summary analysis program (AUTO_STATS) for the month and semester to
date, and the statistics reported to the JCMT Board and in the
Annual Report are essentially these results.
o The analysis performed by AUTO_STATS may be repeated
following identification and correction of spurious entries, or
significant errors in categorization.
o Occasionally reports are not filed on time, and missing reports can
seldom be reconstructed. For example, the reporting for Semester 96B
apparently misses about 8% of shifts, although this incompleteness may be
a result of
the incomplete reconstruction of the database following the reported
disk crash of 25th December 1996.
Last Modification Date 1997/04/22 - Last Modification Author: Graeme Watt (gdw)
|
