Operations for acquiring guide stars and tracking moving targets are described in this article, in the context of JWST's event driven scheduling paradigm. Targets moving up to 75 mas/s can be tracked by JWST with pointing stability comparable to that for fixed targets.

Guide star selection and track generation in the ground system

Moving target guide stars and APT

The handling of JWST guide stars for moving (Solar System) targets in APT differs from that for fixed targets. APT does not look for available guide stars for moving targets when the Visit Planner is run (it is too resource intensive). Instead, guide stars are selected after submission (and acceptance) of a proposal, by the program coordinator. APT does perform one basic check, however. Because it knows the rate of motion of the target vs. time (after the Visit Planner is run), it can estimate the time a typical guide star will remain in the guider field of view. This duration is termed the usable guide star duration.

Depending on the length of a visit (recall that a visit is defined based on the use of a single guide star), the length of the usable GS duration may be shorter than the visit being planned. When that happens, APT issues a warning. If the visit duration is only slightly longer than the usable GS duration, the observation can probably be scheduled successfully. For larger excesses of visit duration relative to usable GS duration, proposers are strongly encouraged to split the observation to produce visits shorter than the usable GS duration. Both the usable GS duration and visit duration are reported in the visit element under the observation element in the tree editor on the left hand side of the APT GUI.

Guide star selection

JWST's scheduling is event-driven (see JWST Observing Overheads and Time Accounting Overview). As a result, the exact time when any observation will execute is not known, even after the weekly schedule is uplinked. For moving targets this means that their position at the time an observation actually executes is not known exactly at the time of scheduling, nor is it known which guide stars will be in the Fine Guidance Sensor (FGS) field of view (FOV) at the time of observation. To allow for this, multiple guide star (GS) candidates are selected for each moving target observation.

The usability window for a given star is determined from the target position as a function of time, and (simplifying somewhat) the size of the FGS FOV. Candidate guide stars are selected with usability windows beginning well before the scheduled start time of an observation, and extending to well past the end of the scheduled time. The JWST flight software chooses the appropriate GS from the candidates at the actual time of execution of the observation. 

Guide star ephemeris generation

Using the target's orbital elements specified in APT and the apparent positions of the guide stars (GS) and the moving target as a function of time, the ground system generates 5^th-order polynomials that describe the path of each GS across the field of view of the FGS. The calculation is such that the moving target will remain stationary in the science instrument reference frame (modulo dither offsets) while the GS moves in the FGS. In essence, the apparent motion of the moving target is compensated by forcing the GS to undergo a reciprocal motion in the FGS. Because knowledge of the position of JWST is somewhat uncertain ahead of time, the GS ephemerides are generated no earlier than about two weeks prior to the scheduled moving target observation, thus minimizing the pointing uncertainty introduced by the uncertainty in the observatory's orbit. 

Guide star acquisition and pointing operations for moving targets

Once a moving target observation begins, flight software selects the 1st candidate GS with earliest and latest start times bracketing the current time (GS 2 in the context of Figure 1). The observatory slews to place that guide star in the FGS field of view and performs guide star identification (see Figure 2).  Flight software then computes and executes the slew, placing the science target at the appropriate location (ambush point) in the science instrument field of view. A small extra amount of time is included such that the slew to the ambush point is guaranteed to complete before the science target arrives there. That padding also allows for a small pointing correction if the GS is slightly displaced from the ambush point location.

During an observation the FGS measures the GS position every 64 ms. Flight software compares that to its pre-computed ephemeris position in the FGS and updates the observatory pointing to force the star to follow the track specified by the polynomials for that GS. The high cadence guiding information is recorded in the level-1b "_uncal" files and used to reconstruct the target's motion across the sky. The heliocentric and observer distances are also included in those FITS headers.

Measured on-orbit tracking performance for moving targets yields pointing stability in the co-moving frame of the target that is comparable to that for fixed targets. The system allows for dither and mosaic offsets for moving targets (not shown in Figure 2), providing exceptional image quality even in level-3 data products for observations including such offsets.

Background sources in moving target observations

Trailing of fixed sources in moving target observations

Because the observatory is slewing while tracking a moving target, all fixed background sources are trailed across the detectors during exposures. This trailing does not degrade pointing stability performance on the moving target because the fixed sources do not get trailed across the (also fixed) guide star in the FGS.

Background sources generate transient signals during integration ramps as they are trailed from pixel to pixel during an integration. The pipeline jump step flags those transients as "jumps" (aka cosmic rays), with the result that much (but not all) of the signal from background sources gets deleted from the level-2 and later products. In general, the core of the PSF of background sources is detected at a SNR high enough to trigger flagging during jump. Lower SNR wings of the PSFs of such sources frequently do not get flagged, however, and therefore appear in the higher level data products as ring-like structures spread along the direction of the moving target's motion vector, as show in Figure 3.

Shadow observations

Observations of moving targets, particularly faint extended sources such as comets, can benefit from "shadow observations." A JWST moving target shadow observation consists of two visits, one taken while tracking the science target (the science visit), and a second taken after the target has moved away (the shadow visit). During the shadow visit the observatory tracks exactly as it did during the science visit, with instrument operations also exactly replicated at the same points along the track. The shadow visit thus provides clean images and/or spectra of the sky that can (in principle) be directly subtracted from the data from the science visit.

Planning shadow observations

Planning a shadow observation in APT is straightforward. An observation of a target is created as usual, resulting in the creation of the science visit. When a Moving Target Shadow Special Requirement is added to the observation, APT automatically creates the duplicate shadow visit and a duplicate target with "-shadow" appended to the name. Note that Moving Target Shadow is found with other general special requirement options, not under the Solar System tab. The special requirement has a parameter, Min Interval, which should be set to no less than the default value of 10 days. That interval guarantees that the actual start time of the shadow visit can be retrieved from telemetry and used by the scheduling system on the ground to fully specify the shadow visit. The 10-day minimum interval will typically be more than sufficient to guarantee that the target and any extended emission will no longer be present when the shadow visit executes, but it is up to the observer to verify that and increase the Min Interval if needed.

Scheduling and execution of shadow observations

Once the science visit has executed, the ground system creates the shadow visit using the same guide star as the science visit. The polynomials specifying the guide star track in the FGS are exactly the same, but the value of the time used in evaluating the polynomials is offset using the current observatory time and the known start time of the science visit. Early on-orbit testing has shown that the track is repeated with a precision of about 0.1", which is sufficient to support shadow subtraction for imaging observations or those obtained with the NIRSpec IFU or MIRI MRS. The same testing also indicates that time used for instrument re-configurations (e.g., filter or grating changes) is very repeatable, and so does not appreciably degrade the utility of data from a shadow visit for background subtraction. 

Another unique feature of shadow observations is that an extra exposure is taken at the beginning of each visit. The exposure is as short as possible, but utilizes the same instrument optical configuration as the first exposure specified in APT. These "set-up exposures" are critical for guaranteeing that the relative timing of exposures along track is exactly repeated for both visits. 

Guiding limitations for moving targets

See also: JWST Guide Stars

The faintest guide stars that can be used for moving targets are ~1 mag brighter than the faintest that can be used for fixed targets. The smaller number of available guide stars does not usually limit schedulability. However, in rare circumstances involving observations with very tight constraints, suitable guide stars may not be available within the constraint windows. Loosening geometric or timing constraints may be required to allow scheduling in rare cases.

The visit splitting distance is 38" for moving targets. APT will automatically split observations into multiple visits when dither or mosaic offsets specified for the observation cause larger motions. It is important to note, however, that the apparent non-sidereal motion of a Solar System target is not accounted for when APT automatically splits visits. That said, the apparent motion of the target is accounted for during guide star selection (see above), so any situations where the approximate visit splitting algorithm in APT is incorrect should be flagged during actual scheduling of the observation. Because the apparent motion of targets observed with JWST is generally much smaller than 38" during an observation, this limitation has not resulted in scheduling or operational issues as of this writing. 

The maximum apparent rate of motion that can be tracked with JWST is in excess of 100 mas/s. While this has been demonstrated on-orbit, data quality is likely to be somewhat degraded at such high rates. For example, dither offsets may not be correctly executed leading to a loss in total exposure time and therefore SNR. The default limit on the track rate applied by APT is 75 mas/s. Proposers should contact the Help Desk if observations at higher track rates are being contemplated. While APT does not directly report track rates for moving targets, it is fairly straightforward to retrieve them using the JPL Horizons tool or astroquery Python library.

References

Milam, S., et al. 2016, PASP, 128, 959 
The James Webb Space Telescope’s Plan for Operations and Instrument Capabilities for Observations in the Solar System
ADS  arXiv