4.4 HST Pointing Accuracy and Stability

Pointing  

An understanding of HST's pointing stability and the accuracy of commanded offsets is essential for planning dither observing strategies, regardless of whether integer or subpixel shifts are desired. Multiple, dithered exposures of the same target with HST could have three types of observing scenarios.

Within a single orbit:

• The pointing stability of HST during the orbit, specifically when pointing at a single location
• The precision with which HST can be offset to different dither locations during an orbit (i.e., a comparison between the commanded and actual tele- scope offsets)

Within a single visit (i.e., multiple contiguous orbits):

• The pointing repeatability after the guide stars are re-acquired at the start of each new orbit

Across multiple visits:

• Whether or not the same guide stars are used
• Repeatability of pointing and roll angle after a full guide star acquisition

Astrodrizzle-combined images from the pipeline are observations taken within a visit (the first and second scenario described above), as specified in the exposure logsheet either using the CR-SPLIT, or special requirements, or dither patterns. Such exposure groupings are known as associations. In the pipeline, exposures in an association are aligned based on the WCS of each image, and drizzle-combined to create a combined distortion-free image with suffix drz.fits. In most of these cases, the image alignment based on WCS, performed by AstroDrizzle in the pipeline, produces satisfactory results. On some occasions, the alignment may not be optimal due to pointing anomalies. These associations will have to be re-processed by the observer who will need to change the default AstroDrizzle processing for an image association to produce better results.

In post-pipeline image processing, AstroDrizzle and other associated tasks can be used for more complex image combination, such as combining observations taken across several visit or taken at different orientations, as described above. Statistics on HST pointing behavior have been continually improving thanks to extensive use of dithering to optimize scientific output in several large observing campaigns. These include the Hubble Deep Field North and South (Williams et al. 1996, 2000; Casertano et al. 2000; Gardner et al. 2000), the Hubble Ultra-Deep Field (Beckwith et al, 2006), and long-term monitoring campaigns of the globular clusters M22 and 47 Tuc (programs 7615 and 8267 respectively; Sahu et al. 2001; Gilliland et al. 2000). These observations have provided an excellent body of information regarding precision and repeatability of HST offsets, as well as tracking stability of the telescope when no offsets are commanded (e.g., multiple exposures at the same location). Drawing on experience with these observing programs, more details about HST pointing and stability characteristics for each of the observing scenarios listed above can be described, particularly in terms of positional accuracy of the spacecraft when performing offsets for dithered observational programs. Gilliland (2005) contains a thorough analysis of the datasets, which established the values given in Table 4.4.

Table 4.4: Typical HST pointing and stability characteristics

Tracking Stability

During each orbit, thermal variations in the telescope cause structural variations known as breathing, which leads to changes not only in the optical telescope assembly (OTA), but also in the way that the Fine Guidance Sensors (FGS) track guide stars. The breathing manifests itself as time-dependent changes in the shape and centroid of the PSF across the image due to the changing focus.

Changes related to the FGS, on the other hand, depend largely on whether fine lock has been achieved on one or two guide stars. Most observations are obtained with successful fine lock on two guide stars. In those cases, positional drifts would mainly be related to thermal variations that predominantly manifests as positional translations. A small amount of rotation may also occur during an orbit, typically less than a few hundredths of a pixel across the science instrument. Typical RMS tracking accuracy is generally on the order of two to five mas or less throughout an orbit, and can be verified post-facto by examining the jitter files that are part of the archival dataset for a particular observation.

In some observations, however, fine lock is successfully achieved on only one guide star. In this case, a steady roll angle drift is present as a result of gyro drift. The telescope will rotate about the guide star, typically at a rotation rate of ~1.5 mas/sec but rates of up to five mas/sec have occurred on rare occasions. This manifests itself primarily as a translation of the science instrument, but some slight rotation may also be evident. The actual amount of translation of the science instrument on the sky depends on its location in the focal plane relative to the guide star. For example, STIS and NICMOS are located approximately midway between the optical axis and the FGS apertures, so their distance from a guide star could range from 6 to 20 arcminutes. For these instruments, the maximal scenario of a rotational drift of five mas/sec would produce a total translation during one orbit ranging ~25 to 85 mas. For WFPC2 this maximum shift could be ~50 mas.

Thus, before proceeding with the analysis of dithered data, it is always advisable to examine either the EXPFLAG keyword value or the jitter data products to confirm whether a two-FGS fine lock was successfully achieved during the observation. If a two-FGS fine lock was achieved, the expected translational shifts due to FGS drift should be less than three mas during the orbit, and any apparent rotation should be less than a few hundredths of a pixel across the detector. The HST Data Handbook contains details on how to extract the relevant information from jitter files.

Precision of Commanded Offsets

If the primary reason for dithering is to avoid bad pixels or improve PSF sampling, then dither offsets less than about one arcsecond are recommended. Examination of HST behavior in previous dither campaigns reveals that for offsets of this size, the actual measured dither offset will agree with the commanded offset to an RMS within about two to five mas during a single orbit with good lock on both guide stars. The RMS of this offset typically increases to a range of up to ~10 to 15 mas when comparing one visit to another over several days. Occasionally, the actual offsets can differ substantially from the commanded offsets by ~0.1 to 0.2 arcseconds or more, with field rotations of up to 0.1??. This is generally the result of FGS false lock on a secondary null, or other FGS interferometric peculiarities. This behavior was observed in two out of nine pointings during the HDF-N campaign.

In some cases, larger dither offsets of up to a few arcseconds are required to bridge inter-chip gaps between detectors, as in WFPC2's four CCDs and the two detectors in ACS/WFC. Offsets of this size are unlikely to present any problems with pointing precision but observers should be aware that such offsets may introduce more non-uniform subsampling across the field as a result of the geometric distortion inherent in the instruments.

Offsets larger than several tens of arcseconds may result in guide stars moving out of the FGS apertures, depending on the exact configuration of the primary and secondary guide stars. This would necessitate a full target acquisition using new guide stars, with substantial associated overhead, as well as a loss of pointing repeatability due to the relative positional uncertainties in the guide star catalog (~0.2 to 0.5 arcseconds). Such large offsets are more appropriate for mosaicing programs where large areas are being mapped, and would therefore involve a fundamentally different proposal design than those programs involving small dither offsets.

Pointing Repeatability After Guide Star Re-acquisition

For many HST programs, dithered observations of a target are obtained during a number of separate orbits, often contiguous, which are in turn grouped into one or more visits.
The first orbit in a visit begins with a full guide star acquisition. For each subsequent orbit in the same visit, HST will reacquire the same guide stars upon exit from occultation. In post-occultation guide star reacquisitions, the instrument pointing is typically within ~5 to 20 mas of its location in the previous orbit.

The precision of HST's guide star reacquisition is based on its ability to force the post-slew position of the guide stars to reside in the exact same location in the guide star acquisition field-of-view (i.e., pickles), as in the previous orbit. This is generally sufficient to reliably perform subpixel dithers for most HST instruments that have pixel sizes of the order ~0.05 to 0.1 arcseconds. Therefore, it's recommended that, whenever possible, the observing proposal should be designed to fit all dithered observations of a given target into a contiguous set of orbits within a single visit to provide improved relative image registration.

Roll Angle Repeatability Over Multiple Visits

Some observing programs are sufficiently large enough to necessitate dithered observations of the same target over many orbits. In such cases, it is necessary to break the observations into several visits because the length of a single visit is constrained by available scheduling windows depending on the target's position in the sky. For all targets outside the CVZ, single visits are usually constrained by scheduling limitations to contain no more than five orbits. If multiple visits of the same target were scheduled across different dates, images in one visit may have small offsets relative to images from other visits, even if the same pointing, same guide stars, and same ORIENT were specified for the visits.

At the start of a new visit, HST sets up the specified roll for the observation using the gyros, and carries out a full acquisition of the dominant guide star. This is followed by the acquisition of the sub-dominant guide star, which enables the telescope to track in fine lock. The pointing control system (PCS) then preserves this roll angle for the remainder of the visit.

In most cases, the difference between the desired roll angle, and the actual roll angle, will be less than ~0.003??. This corresponds to a positional shift of approximately 73 mas at the sub-dominant guide star, assuming a separation of 1,400 arcseconds between the two guide stars. For WFPC2, this shift is 38 mas, i.e., just less than the size of a WFPC2/PC pixel. Therefore, multiple visits at the same specified roll, target, and guide stars will, under nominal circumstances, show repeatability to this level. It is not uncommon for scheduling constraints to affect the time between updates to the Fixed Head Star Trackers (FHSTs) and FGS acquisitions, in which case roll angle deviations of 0.01 ??and greater can occur (i.e., translational shifts above 100 mas). For visits with the same guide stars and requested roll angle, the actual roll changes incurred between visits can be accurately determined from the locations of guide stars in the FGS as recorded in the datasets' jitter files.