Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/hst/stis/documents/isrs/199623.pdf Дата изменения: Tue Dec 7 08:04:01 2004 Дата индексирования: Sat Dec 22 23:30:47 2007 Кодировка:

STIS Instrument Science Report 96-023

STIS Pipeline Testing II: TIMETAG Processing
S.R. Keener, S. Hulbert and N. Zarate May 1996

ABSTRACT This report summarizes the testing of the processing of TIMETAG data by the task, inttag. The inttag task converts a table of events into a 2-D image suitable for further processing. Problems noted during the testing were corrected. The final testing confirmed nominal performance of the inttag task, except for three open issues which will be addressed in future reworks of the task. The three open issues are the need for a Control-C error handler, process abortion (plus printing of an error message) if the input contains overlapping Good Time Intervals, and the addition of the EXTVER keywords to output image headers.

1. Introduction
The inttag task is a stand-alone task that will convert an EVENTS table of TIMETAG mode STIS data, generated by generic conversion, into a raw, time-integrated, ACCUM image. This image can then be run through other STIS processing and calibration software. The inttag task is specifically designed as a preprocessor to format TIMETAG data such that the data can be processed through CALSTIS. See STIS ISR 96-011 for more information about the commanding and operation of TIMETAG mode and STIS ISR 96-013 for more information about the pipeline processing of TIMETAG mode data.

2. Task Calling Sequence
The inttag task allows the user to input necessary variables on the command line and will prompt the user if the incorrect number of variables was entered. The task will eventually be a stand-alone task within IRAF. During testing, the calling sequence used was:

1


inttag

input output hires/lores start_time time_interval nintervals

When a user wishes to create only one accum image with all input events, the user enters only the first three variables (input table filename, output image filename, hires or lores) on the command line. The inttag variables start_time and time_interval allow the user to specify which interval to integrate over. This interval is checked against the Good Time Intervals (GTIs) for that image. When specifying a start_time and time_interval, a user may also specify the number of accum images to create within the output FITS file by using the nintervals variable. The nintervals images are of the same duration (specified by the time_interval variable) and successively follow one another with no gaps except for those marked by the GTI table. (In STIS ISR 96-013, time_interval and nintervals were called duration and number_of_intervals.)

3. Testing
The first action items in the test plan for the inttag task included making timetag fits tables for testing both spectroscopy and imaging mode observations. The definition of TIMETAG data format was taken from ISR 95-013 and used to create files of the correct format which contained all the necessary keywords. The X and Y event stream coordinates were randomly generated. Some input images also included some zeros in the X and Y columns. Both square and non-square, as well as full size and subarray images were used as input for testing. A timetag fits file with four event extension columns (TIME, X, Y, DETAXIS) was used for testing a "Doppler-corrected" input to inttag. An input fits file with three event columns (TIME, X, Y) was used to test a "non-doppler-corrected" input to inttag. All these tests ran nominally. There were no errors, and the inttag task correctly calculated exposure times including only the GTIs within the desired integration interval. The inttag output file structure was checked. The output file had science, error, and data quality extensions for each integrated image. As expected, the data quality extension was made of all zeros since only events within good time intervals were included. The error extension had nonzero pixels equal to the square root of the number of counts in corresponding science extension pixels. The inttag task correctly copied the full input event table header, which contained all the keywords needed by each extension for further image processing, into each of the output image extension headers. The keywords EXTNAME and EXTVER are supposed to be added to the output image extension headers by the task. However, EXTVER was not. This will need to be corrected in future versions of the inttag task. Several different tests of inttag's integration ability (including correctly applying good time intervals) were run. Data with repeated events at the same locations were cre-

2


ated. The GTIs included some of these repeated events but not others. The inttag task correctly added up only the events at these locations which were within the GTIs. When a lores output was specified, the integration and binning of events within appropriate GTIs was correctly performed. Input files with only one GTI near the beginning of the image exposure, more than one GTI spread out over the exposure time, one GTI which included all the exposure time, one GTI in the middle of the exposure time, one GTI which ended past the last event in the events table, and two overlapping GTIs were used in the testing. All these test cases came out nominal, except for the input file with overlapping GTIs. The inttag task did correctly add up the events in the overlapping GTIs, but when calculating the exposure time for the output image, it counted overlapping time twice. Generic conversion should create only sequentially incremental good time intervals. A future version of the task will need to produce an error message and stop processing if there are overlapping GTIs in the input fits table. Testing was done with values of start_time and time_interval which overlapped and with values which did not overlap some GTIs. When the time_interval variable went beyond the end time of the last GTI, inttag ran without errors and correctly included only events within the appropriate GTIs. If both the start_time and time_interval variables were beyond the last GTI, then an error warning of this was given, and processing stopped. When the interval variables entered were between GTIs, a warning that this was true was printed, and processing stopped. Several values of nintervals were tested with input standard size images and subarray images. The results were nominal. Any data within a GTI that was also within the time_interval was correctly integrated for each of the accum images. The format of the output image included three extensions (SCI, ERR, DQ) for each accum image. For example, the output image for nintervals = 2 had six extensions altogether.

4. Conclusions and Open Issues
Final testing results show that the inttag task functions nominally. However, there are several remaining open issues for the current version of inttag. These issues need to be addressed during future reworks and then tested. An error handler that will allow a user to Control-C out of the task is needed. As is the case with calstis-1 (see STIS-ISR 96-022) when a user uses Control-C to exit the task, error messages are continuously generated without stopping the processing. An error message should be printed and processing stopped when the input fits table contains overlapping GTIs. Finally, the output images produced by the inttag task need to contain the EXTVER keyword in each extension. Testing of the calstis-1 processing of inttag output images can be accomplished when this last issue is resolved.

3