Note: The jitter-correction routines use information from a new file, *jitrf.fit, which contains spacecraft-pointing information. Because jitter files were not generated before August 2002, the jitter-correction routines cannot be applied to older data sets.
Introduction
The term ``jitter'' normally refers to small, rapid variations in the telescope pointing. Until recently, the positioning of FUSE was controlled entirely with reaction wheels, which typically maintain a pointing accuracy of 0.2-0.3 arcsec. These motions resulted in a slight smearing of the spectra. Because this smearing was much less than a resolution element, no attempt was made to correct the spectra for this effect. In late 2001, when FUSE lost the use of two of its four reaction wheels, it became necessary to control the spacecraft motion along one axis with magnetic torquer bars. These torquer bars can exert only about 10% of the force produced by the reaction wheels, and the force available depends strongly on the strength and relative orientation of the earth's magnetic field. Thus, while the amplitude of the jitter remained essentially the same along the two axes controlled by reaction wheels, motion increased dramatically along the third axis, termed the antisymmetric or A axis. The magnitude of the jitter along the A axis increases with decreasing declination and is accompanied by longer-term, orbital phase-related drifts which can exceed several arc seconds. These motions can substantially degrade the resolution of the spectra, so procedures have been implemented to correct the data for spacecraft motion during an observation. For time-tagged observation, we can reposition individual photon events. For for histogram observations, we correct only for exposure time lost to large excursions of the spacecraft.
The Housekeeping File
The heart of the correction procedure is the production of a FITS file containing spacecraft-pointing information tabulated each second during an observation. Called a housekeeping file, it is generated during level-zero processing (LZP; before the CalFUSE pipeline is run) and is archived with the raw data file. Housekeeping files are named with the suffix ``hskpf'' appended to the observation designation. (See the FUSE Instrument and Data Handbook for a discussion of file-naming conventions.) For example, the housekeeping file for observation A1010201001 is called A1010201001hskpf.fit. When this file is generated, the keyword HKEXISTS in the raw data file is set to ``yes.''
The housekeeping file is derived from engineering data supplied by the spacecraft. The most important of these data are the quaternions, four-element vectors that completely specify the orientation of the telescope. The spacecraft produces two estimates of the quaternions. The first is derived from the measured positions of two to six guide stars; these are normally referred to as FPD (for Fine Pointing Data) quaternions. The second combines guide-star data with data from the gyros and magnetometer. Since they are generated by the onboard ACS (Attitude Control System) computer, they are normally referred to as the ACS quaternions. The guide stars are generally more accurate than the gyros and magnetometers, so the FPD quaternions are generally more accurate than the ACS estimates. Occasionally, there are problems with the guide-star measurements. This typically happens if the guide stars are faint or the background against which the guide stars are measured is high (normally around orbital noon). The tabulated guide-star coordinates may also be inaccurate. If this situation occurs, then the questionable guide stars are removed from the quaternion solution. In some cases, guide-star lock is lost completely and the only available quaternion estimates are the ACS values. It is for this reason that the ACS estimates are the ones used to control the pointing of the satellite.
The Jitter File
From the housekeeping file, a separate LZP routine derives a FITS-format jitter file. Also archived with the raw data, the jitter file has the suffix ``jitrf'' appended to the observation designation. It consists of a binary table with 4 columns: TIME, DX, DY and TRKFLG. The time refers to the elapsed time (in seconds) from the start of the observation. Since the engineering data commonly begin up to a minute before the exposure, the first few times are negative. DX and DY are the offsets along the X (dispersion) and Y (cross-dispersion) directions in arc seconds. These offsets are relative to the commanded position of the telescope. Finally, TRKFLG is a tracking quality flag that indicates the reliability of the tabulated positions. A value of 5 indicates that the telescope was tracking on guide stars (i.e., fine guiding) and that the offsets were calculated with the FPD quaternions. A value of 4 indicates that the telescope was not tracking on guide stars (coarse guiding), but that guide star data are available and are being used to generate the positions. A value of 3 means that guide stars were not available and the offsets are based on ACS quaternions. Finally, a value of -999 indiates that no information was available for that time interval.
The top-level FITS header contains a number of keywords that summarize the pointing history during the exposure. These keywords are useful for evaluating the quality of the tracking. A list of the most useful keywords follows.
Keyword Description TIMEOBS UT start time of exposure (hh:mm:ss) EXPSTART Exposure start time (Modified Julian Date) EXPTIME [sec] Exposure duration -- predicted JIT_STAT status of jitter data is good (if = 0) EXP_DUR exposure duration - measured FINE_GDE fraction of exp in fine guide COARSGDE fraction of exp in coarse guide NOGDEIN fraction of exp with no guiding info GS1_USED fraction of exp using guide star 1 GS2_USED fraction of exp using guide star 2 GS3_USED fraction of exp using guide star 3 GS4_USED fraction of exp using guide star 4 GS5_USED fraction of exp using guide star 5 GS6_USED fraction of exp using guide star 6 NGS_USED number of guide stars used KNOWNTRK fraction of exp tracking on known stars UNKWNTRK fraction of exp tracking on unknown stars GS_INUSE fraction of exp tracking on guide stars SLEWFLG slew commanded during obs (if > 0) POSAVG_X [arcsec] mean DX during exposure POSAVG_Y [arcsec] mean DY during exposure X_JITTER [arcsec] sigma of DX during exposure Y_JITTER [arcsec] sigma of DY during exposure X_JIT_5M [arcsec] sigma of DX during last 5 min of exp Y_JIT_5M [arcsec] sigma of DY during last 5 min of exp X_JITLRG frac of DX more than 2-sigma from POSAVE_X Y_JITLR frac of DY more than 2-sigma from POSAVE_Y
The Jitter-Correction Module The CalFUSE pipeline corrects for spacecraft motion if four conditions are satisfied: (1) the keyword HKEXISTS in the raw data file is set to ``YES,'' (2) the target is a point source (the keyword SRC_TYPE begins with ``P''), (3) the keyword RUN_JITR in the corresponding parameter file (parm*.fit) is set to ``YES'' (default), and (4) the status keyword JIT_STAT in the jitter file is set to zero, indicating that the file contains valid data. If these conditions are not met, the program copies input to output, issues a warning message, and exits, setting the keyword JITR_COR to ``SKIPPED.'' If the conditions are met, the program searches the current directory for the jitter file. If it cannot find the file, it exits and issues an error message. This may happen for some data sets processed in the summer of 2002, for which housekeeping but not jitter files exist. For such data sets, we suggest that you set RUN_JITR to ``NO'' in the parm*.fit files.
Finding the jitter file, the program proceeds in two steps: First, it rejects photon events for which we have no or incomplete pointing information. To do that, the program uses two keywords from the parameter file parm*.fit. The keyword J_TRKFLG represents the lowest acceptable value of TRKFLG. For example, a threshold of 3 (default) will cause the program to reject all photon events that arrive during times for which TRKFLG < 3. The other keyword, JTIMEGAP (default = 60), authorizes the program to retain photons arriving during periods with low values of TRKFLG if those periods are less than JTIMEGAP seconds in length. Second, the program applies the jitter correction. For each second during an observation, the corrections DX and DY are read from the jitter file and added to the DX and DY columns of the time-tagged data file. During time periods (less than JTIMEGAP seconds in length) for which TRKFLG < J_TRKFLG, the program estimates the pointing by interpolating between good values of DX and DY. Whenever |DY| > 30 arcsec, the program assumes that the target has moved out of the aperture and rejects all associated photon events.
Note that the shifts in the jitter file are in arcseconds, while the shifts used by the pipeline are in pixels. To scale arcseconds to pixels, we use the following conversion factors:
1A 1B 2A 2B
DX : 1.743 1.743 -1.743 -1.743
DY : 1.154 1.163 -0.633 -0.581
These shifts not applied until the final image file is constructed by
the pipeline module cf_ttag_geodopp.
For Histogram Data
We cannot correct the positions of individual photons events for data obtained in histogram mode. Instead, the program cf_hist_jitter reads the jitter file, looking for periods when the target is out of the aperture (that is, times for which |DY| > 30"), and writes the total time lost to jitter in the header keyword EXP_JITR. Later in the pipeline, EXP_JITR is subtracted from EXPTIME before the extracted spectra are flux calibrated.
Questions?
Please address questions to
fuse_support@pha.jhu.edu.
WebMasters: