Fine Guidance Sensor Instrument Handbook for Cycle 14
 |
 |
 |
 |
 |
|
| Fine Guidance Sensor Instrument Handbook for Cycle 14 | |||||
| ||||||
7.1 Data Processing OverviewChapter 7:
FGS Astrometry
Data Processing
7.2 Exposure-Level Processing
7.2.1 Initial Pipeline Processing
7.2.2 Observing Mode Dependent Processing
7.3 Visit-Level Processing
7.3.1 Position Mode
7.3.2 Transfer Mode
7.4 Epoch-Level Processing
7.4.1 Parallax, Proper Motion, and Reflex Motion
7.4.2 Binary Stars and Orbital Elements
This chapter describes the FGS astrometry data processing pipeline and analysis tools. The steps to reduce and calibrate the raw data are described along with the sequence in which they are applied.
7.1 Data Processing Overview
FGS Astrometry observations are analyzed at three distinct levels, the exposure-level (individual observations), the visit-level (all observations within the HST orbit), and the epoch-level (relating data from one visit to others). The astrometry data pipeline processes the observations up to and including the visit level. Epoch-level analysis requires tools beyond the scope of the FGS data pipeline.
The exposure- and visit-level corrections and calibrations are performed by the observer using calfgsa and calfgsb which are implemented as tasks in STScI's STSDAS system. These tasks are semi-automated and require little user input to process the individual exposures that comprise the typical astrometry visit. Reference files used by these tasks are maintained by STScI and can be found by following the links to the calibration sections of the FGS Web Site at
http://www.stsci.edu/hst/fgs/.Epoch-level analysis of FGS data is not, by its nature, a procedure which lends itself to generic pipeline processing. However, tools to provide the observer with some of the more common manipulations encountered in data analysis of FGS astrometry observations are being made available to the general FGS user. Currently, these tools are not STSDAS tasks, but a collection of stand-alone scripts and executable files to achieve plate solutions for Position mode observations and the deconvolution of binary star transfer functions from Transfer mode observations.
The processing and analysis applied at each level is discussed below. Interested readers are encouraged to monitor the STScI newsletter or visit the FGS web site for updates to the status of these tools. More detailed discussions can be found in the
HST Data Handbookversion 4.0 or later.7.2 Exposure-Level Processing
The term "exposure-level processing" refers to pipeline corrections that are applied to the individual FGS observations. These are discussed in this section.
7.2.1 Initial Pipeline Processing
Regardless of the observing mode, several activities are carried out during the initialization of the astrometry pipeline. This begins with the usual file management, data quality assessments, and the determination of the required reference files and their availability status. At this early stage, the data are inspected to determine the identification of the astrometer FGS, its mode of operation, and the availability of guide star data from the guiding FGSs. The astrometer's data are inspected to evaluate the outcome of the Search, CoarseTrack, and FineLock target acquisitions, while the guide star data are inspected to identify the guiding mode (i.e., was the spacecraft guided by one or two guide stars, and were the guide stars tracked in FineLock?).
If the astrometry target acquisition failed, the FGS flags and status bits are inspected to determine the cause. In this case, data processing proceeds as far as possible (in the event that the observation was a partial success), output files are generated and populated appropriately, and pipeline processing of the observation terminates.
7.2.2 Observing Mode Dependent Processing
After pipeline initialization and data quality assessments, successful observations (i.e., those that acquired the target), are processed according to the FGS observation mode: Position or Transfer.
Position mode
The goal of exposure-level Transfer mode pipeline processing is to determine the centroid of the IFOV while the FGS tracked the object in FineLock. A collateral objective is to analyze the individual PMT data, both to determine the small angle corrections that need to be applied to the centroids as well as to provide photometric information about the target and the sky background.
The guide star data are analyzed in the same way as the astrometer data, over the identical intervals of time. For example, the guide star centroids and average photometry are computed over the time the astrometer was in FineLock.
The corrections applied to the FGS data are as follows:
- The FineLock centroids are computed by finding the median, from the 40 Hz data - of the X,Y location of the IFOV (computed from the Star Selector A,B encoder angles). PMT data are averaged for astrometer and guiding FGSs.
- For the astrometer only, the PMT data are evaluated to determine the fine angle adjustments to the centroids.
- The Optical Field Angle Distortion (OFAD) calibration is applied to remove distortions of the sky in the FOV.
- Differential velocity aberration correction is applied to the adjusted FineLock centroids of the astrometer and guiding FGSs.
Steps 1 and 2 are carried out in calfgsa, while steps 3 and 4 are performed in calfgsb. Please see Figure 7.1 and Figure 7.3, the flow chart descriptions of calfgsa and calfgsb respectively.
Transfer mode
During a Transfer mode observation, the data retrieved from the astrometric FGS will include PMT counts and star selector positions from the slew of the IFOV to the target object, the Search and CoarseTrack target acquisition, and the individual Transfer scans. Corresponding data acquired from the guiding FGSs will include FineLock tracking of the guide stars.
The astrometer's data are analyzed to evaluate the background counts, if available (see Chapter 7), and to locate and extract the individual scans. For each scan, the guide star centroids are computed and corrected for differential velocity aberration. Output files are generated with the appropriate information.
The data from the individual scans are used to compute the Transfer Function over the scan path. The quality of each scan is evaluated for corruption from high amplitude, high frequency spacecraft jitter, and, if unacceptably large, the scan is disqualified from further analysis.
The remaining scans are cross correlated, shifted as needed, binned as desired, and co-added to enhance the signal to noise ratio. The co-added Transfer Function can be smoothed if need be. The analysis tool which performs these functions is available from STScI. It is currently implemented as a standalone executable (FORTRAN + C) in the UNIX environment.
Although not part of pipeline processing, the analysis of observations of binary stars and extended objects will be briefly described here for completeness.
Binary Star Analysis
The Transfer Function of a binary system will be deconvolved, by use of the standard reference S-Curves of single stars from the calibration database, into two linearly superimposed point source S-Curves, each scaled by the relative brightness and shifted by the projected angular separation of the binary's components. This provides the observer with the angular separation and position angle of the components as well as their magnitude difference. These results will be combined with those obtained from observations at different epochs to compute the system's relative orbit.
Extended Source Analysis
For observations of an extended source, such as the resolved disk of a giant star or solar system object, the co-added Transfer Function will be analyzed to determine the angular size of the object. This involves application of a model which generates the Transfer Function of synthetic disks from point source S-Curves from the calibration database.
Transfer mode observations are processed by calfgsa to the point of locating and extracting the individual scans and computing the guide star centroids. Support for additional processing - including the automation of the data quality assessment (identifying those scans which have been unacceptably corrupted by space craft jitter, for example) and the cross correlation and co-adding of the individual scans - are available as data analysis tools. Upgrades to calfgsa will be noted on the FGS web site. Figure 7.2 displays the processing steps performed by the current version of calfgsa for Transfer mode observations.
7.3 Visit-Level Processing
Visit-level processing refers to those corrections that are applied to the individual exposures in order to map each onto a common reference frame. Since the FGS observes the targets sequentially, not simultaneously, any motion of the spacecraft or the FGS's FOV during the course of the visit will introduce uncertainties in the measured positions of the objects. The corrections discussed here restore the cohesiveness of the reference frame.
7.3.1 Position Mode
Position mode observations during the course of a visit must be corrected for two sources of error which render the FOV somewhat unstable: low frequency HST oscillations and residual drift of the FOV across the sky.
HST Oscillations: Using the Guide Star Data
The mapping of the individual astrometry observations onto a common reference frame begins with the analysis of the guide star data. As part of the exposure-level processing, guide star centroids are computed over the same time interval as for the astrometry targets. Generally HST is guided by two guide stars. The so-called "dominant" guide star is used by the pointing control system (PCS) to control the translational pitch and yaw of the telescope. The "sub-dominant" guide star, also referred to as the roll star, is used to maintain the spacecraft's roll or orientation on the sky. Any change in the guide star centroids over the course of the visit (after corrections for differential velocity aberration) is interpreted by the FGS Astrometry Pipeline as an uncorrected change in the spacecraft's pointing.
The pipeline defines an arbitrary fiducial reference frame based upon the location of the guide stars in the first exposure of the visit. Relative changes in the position of the dominant guide star for subsequent observations are assumed to be a translational motion of the HST focal plane. The pipeline "corrects" the position of the sub-dominant guide star and the astrometric target star. Any change in the angle defined by the line connecting the two guide stars and the spacecraft's V2 axis is interpreted as a rotation of the focal plane, and is removed from the astrometry data.
The correction of the astrometry centroids for vehicle motion (as determined by changes in guide star positions) is referred to as pos-mode dejittering. Transient corrections can be as large as 3 to 5 mas, such as when HST enters orbital day, but the adjustments are typically small-less than 1 mas. This underscores the excellent performance of HST's pointing control system under the guidance of the FGSs.
Drift Correction
"Drift", as discussed in Chapter 5, is defined as the apparent motion of the astrometer' s FOV on the sky during the course of the visit as detected by the astrometry targets that are observed more than once during the visit (the check stars). Drift must be removed from the measured position of all astrometry targets. This is accomplished by using check star data to construct a model to determine the corrections to be applied. If at least two check stars are available and were observed with sufficient frequency (i.e., at least every seven minutes), a quadratic drift model (in time) can be used to correct for both translation and rotation of the FOV. The availability of only one check star will limit the model to translation corrections only. If the check stars were observed too infrequently, then a linear model will be applied.
If no check stars were observed, the drift cannot be removed and the astrometry will be contaminated with positional errors as large as 15 mas.
It is important to note that the drift is motion in the astrometer which remains after the guide star corrections have been applied. Its cause is not well understood, but with proper check star observing, the residuals of the drift correction are tolerably small (i.e., sub-mas).
With the application of the guide star data for pos-mode dejittering and the check stars to eliminate drift, the astrometry measurements from the individual exposures can be reliably assembled onto a common reference frame to define the visit's plate.
The visit level corrections to Position mode observations, i.e., pos-mode de-jittering and the drift correction are performed by calfgsb (Figure 7.3).
7.3.2 Transfer Mode
Transfer mode observations typically last about 20 minutes (or more), much longer than Position mode exposures (1 to 3 minutes). Therefore, it is far more likely that low frequency spacecraft jitter and FOV drift will have occurred during the Transfer mode exposure. These do not introduce uncorrectable errors since low frequency FOV motion is implicitly removed from the data by cross correlating the Transfer Function from each individual scan. However, relating the arbitrary coordinate system upon which the Transfer Function is mapped to the system common to the reference stars is an important and necessary prerequisite in linking the Transfer mode observation to the Position mode data.
Guide Star Data
Transfer mode data analysis, as discussed in the exposure level section, involves the cross correlation of the Transfer Functions from each of the individual scans. The first scan is arbitrarily designated as the fiducial; all other scans in the Transfer Function are shifted to align with that of the first (this automatically eliminates jitter and drift local to the observation). Therefore, in order to restore some level of correlation with the other observations in the visit, the guide star centroids are evaluated over each scan, and, along with the shifts, are recorded.
Drift Correction
The cross correlation of the individual scans removes the drift of the FOV from the Transfer mode data. However, this is a relative correction, local to only the Transfer mode observation. By recording the shift corrections applied to the individual scans, the visit level pipeline has visibility to the drift that occurred during the Transfer mode observation.
Transfer/Position Mode Bias
The presence of a small roll error of the Koesters prism about the normal to its entrance face (see Transfer Mode Scale as a Function of HST Roll Angle) introduces a bias in the location of interferometric null as measured by Position mode when compared to the same location in Transfer mode. This bias must be accounted for when mapping of the results of the Transfer mode analysis onto the visit level plate defined by the Position mode measurements of the reference stars. This bias is removed by applying parameters from the calibration database.
7.4 Epoch-Level Processing
Astrometry takes time. This is true whether the goal is to determine the parallax, proper motion, or reflex motion of an object measured in Position mode, or the orbital elements of a binary system observed in Transfer mode. By its very nature, astrometry looks for changes to the arrangement of objects on the sky, and as a result, observations taken over several different visits must be compared to one another. Relating the observations from different epochs is discussed. A more detailed discussion is provided in then
HST Data Handbook.7.4.1 Parallax, Proper Motion, and Reflex Motion
In order to measure the proper motion and parallax of an object observed in Position mode with the FGS, the data from the individual visits must be combined to form a virtual plate. This virtual plate is derived from an optimal mapping of all of the visit level plates onto a common plate using the method of least-squares to minimize the residuals of all reference stars (the plate solution). This mapping function is used to map the science target at each visit onto the virtual plate. In this way, the parallax, proper motion, and perhaps reflex motion (perturbations caused by a gravitationally bound companion) can be determined.
If enough reference stars are available (> 5), six parameter plate solutions - allowing for independent scale adjustments along each of the astrometer's X and Y axes - can be applied. Otherwise the four parameter model must be applied.
The residuals of the reference stars in the plate solution determine the overall astrometric performance of the telescope as a function of the number of visits expended. FGS3 achieved ~1.2 mas (rms) precision per HST orbit, while FGS1r has demonstrated ~0.8 mas rms precision per HST orbit. Given that the overall astrometric accuracy scales as
while random (Poisson) errors dominate, it can be anticipated that, for example, FGS1r will yield parallax measurements accurate to 0.3 mas in as little as 12 HST orbits if an optimal observing strategy is employed. (Below ~0.2 mas irreducible systematic errors dominate, such as the conversion of relative to absolute parallax.)
STScI can provide the analysis tools (developed by the STAT at the University of Texas) needed to perform a plate solution from multiple Position mode visits. These tools are available as stand-alone software packages and scripts that can be delivered via ftp from STScI. Supporting documentation is being developed. Please see the FGS web page for further updates.
7.4.2 Binary Stars and Orbital Elements
Transfer mode observations of binary stars provide the angular separation and position angle of the components at each epoch (as well as their difference in brightness). Once the binary has been observed at a sufficient number of phases in its orbit, the system's orbital parameters can be solved. The result will be knowledge of the orbit's inclination, eccentricity, period, and angular extent of its semi-major axis. If the parallax of the object is known, then the physical size of the orbit can be computed to yield the total mass of the system.
For a binary that was observed along with Position mode observations of reference stars, it is possible to determine the binary's parallax and proper motion, and the motion of the components about the system's barycenter, which then yields the component masses. This, along with the differential and system photometry (also measured by the FGS), provides the mass-luminosity relation.
STScI can provide observers, via ftp, with the appropriate analysis tools needed to analyze Transfer mode observations of binary star systems. These tools are standalone software packages written by STScI based upon algorithms developed by the STAT at the Lowell Observatory. These packages allow one to deconvolve the binary star Transfer mode data in order to determine the angular separation and relative brightness of the components at each epoch of observation. Documentation of these packages will be made available during Cycle 14. Check the FGS Web pages for updates.
Figure 7.1: CALFGSA Common Processing Tasks
Figure 7.2: CALFGSA Transfer Mode Processing Tasks
Figure 7.3: CALFGSB Position Mode Processing
|
|||||
|
Space Telescope Science Institute http://www.stsci.edu Voice: (410) 338-1082 help@stsci.edu |
