11.2 Exposure Sequences and Contemporaneous Calibrations

There are several instances when a series of associated STIS exposures (rather than a single exposure) will be taken. The data from these exposure sequences are generally processed as a single unit through the STScI calibration pipeline, with the scientific data from the multiple associated exposures appearing in a single file (for a high-level overview of the STIS calibration pipeline and the data product format see Chapter 15). While it is not needed to specify the planned series of associated exposures in the Phase I proposal, it is helpful to know about these sequences when planning the proposal. Once the proposal has been accepted, it is possible to see and use these sequences to prepare the observations during the Phase II. All the information needed to schedule the observations are generated from a single exposure logsheet line in the Phase II proposal.

We discuss several types of associated exposures below:

• Automatic wavecal exposures taken with scientific data to allow calibration of the spectroscopic and spatial zero points.
• CCD CR-SPLIT exposures taken to allow removal of cosmic rays in the scientific data during post-observation data processing.
• Multiple identical repeat exposures, which can be taken to provide time intervals of tens of seconds (CCD) or minutes (MAMA).
• Pattern sequences, in which the target is stepped, for example along the slit to mitigate the impact of hot pixels or perpendicular to the slit to map a two-dimensional region for spectroscopic observations, or in a dither pattern for imaging observations.

In addition, there are two types of contemporaneous calibration observations that observers may take with their scientific observations in special circumstances:

• GO wavecals, taken if exceptional wavelength accuracy is required, or for slitless spectroscopy.
• CCD fringe flats (CCDFLAT) which need to be taken for near-infrared (NIR; λ > 7500 Å) observations in the grating modes if high ­signal-to-noise is required.

11.2.1 Auto-Wavecals

The STIS optical path from source to detector passes through the aperture (slit) wheel (where the filters for imaging also are housed) and then reflects from one of the elements mounted on the Mode Selection Mechanism (MSM) which houses the first-order gratings, the prism, the cross-dispersers for use with the echelles, and the mirrors for imaging work (see Figure 3.1). Lack of exact repeatability in the MSM causes the center of the spectrogram (as defined by the aperture and wavelength centers) to fall on a slightly different detector location each time there is a movement of the MSM (the MSM-induced offsets in dispersion and the spatial direction have been measured to be ±3 pixels or less). In addition, for MAMA first order spectrographic observations, the aperture location on the detector is deliberately shifted each month to ensure equalization of extracted charge across the detector.

To allow calibration of the zero point of the aperture location and the zero point of the wavelength scale for spectroscopic observations, a line lamp observation (so-called wavecal) is taken automatically each time the MSM is moved. In addition, if a series of exposures or a single long exposure is taken at a single MSM setting, then an additional wavecal will automatically be taken when there is a pause in data taking if 40 minutes of exposure time has passed since the previous wavecal. Here, 40 minutes is the time constant for thermal changes which might affect the wavelength accuracy. Testing in orbit has shown that in extreme conditions (when there is a swing from hot to cold), worst-case thermal shifts of roughly 0.3 pix/hr can be seen; however, monitoring shows that under typical observing conditions, thermal drifts are of the order of 0.1 pix/hr (see also the Monitoring page on the STIS website).

To summarize, each set of spectroscopic scientific exposures taken at a given grating tilt (i.e., MSM position) will normally be accompanied by at least one automatically taken wavecal exposure. This auto-wavecal will be taken prior to the science exposure. Additional auto-wavecals will be taken at the first pause in data taking after ~2300 seconds of exposure time at that grating position have elapsed. For MAMA observations in particular, auto-wavecals can be forced into occultation by (1) keeping exposures in an orbit at the same grating setting, and implementing either (2a) a single long exposure that fills the visibility period or (2b) shorter exposures whose sum is less than ~2300 seconds since the last auto-wavecal, followed by an exposure that crosses the 2300 second boundary and fills the remainder of the visibility period. Similarly, if spectroscopic drifts are a concern, MAMA observations can be split into exposure sums that are ~2300 seconds in duration to maximize the occurrence of auto-wavecals. Multiple auto-wavecals occur more naturally for CCD observations which are usually CR-split into exposures of ~1000 seconds, or less, in duration.

These wavecal exposures will be processed along with the scientific data, and they will be used by the pipeline to automatically correct the zero-point offsets in the wavelength and spatial scales (see Chapter 15).

The automatic wavecals are designed to be of sufficient duration to produce spectrograms which contain at least 3 emission lines with 3 counts/pix and 50 counts summed over the line. In those regions of the spectrum where 3 lines are not obtainable, there will be at least 1 emission line with 18 counts/pix and 300 counts summed over the line. For the CCD, where integration times are short, the auto-wavecals will typically be taken to ensure roughly 8 times this signal.  While the fading of the calibration lamps has reduced the count rates over time (particularly at the shortest wavelengths), the default auto-wavecals still appear to provide adequate zero points for nearly all combinations of grating, central wavelength, and aperture.  Questions regarding particular configurations may be addressed to the STIS team via the HST Help Desk.

The combination of thermal changes between the wavecal and scientific exposures, coupled with the ability to measure the zero points in the wavecal exposures, limits the accuracy of the absolute zero points to ≤0.2 pixels (see Section 16.1). Observers who desire more accurate wavelengths than those that will be automatically provided or who are particularly concerned about the time variation of the zero point may also take their own wavecal exposures, using the WAVE target option (see GO Wavecals), in addition to (or instead of) the default auto-wavecals.  The auto-wavecals can be omitted by using the exposure level optional parameter WAVECAL=NO. Note, however, that using this option on any exposure in a visit will turn off the auto-wavecals for all exposures in that visit. The observer will then have to include sufficient GO wavecals to allow adequate wavelength calibration of their data.  Beginning in Cycle 25, the WAVECAL=NO optional parameter was included as a supported availability mode, so that observers no longer need approval to disable auto-wavecals to reduce overhead during orbital visibility. Observers will be required, however, to include GO wavecals at each MSM position when using this option, and will need approval to use longer exposure times than the default values for their GO wavecals.

GO Wavecals

Only if Observers require particularly accurate wavelengths, they can use the TARGET=WAVE option to insert additional wavecal exposures into the observing sequence.

The wavecals taken with TARGET=WAVE are identical to those taken automatically (i.e., the auto-wavecals) with two important exceptions. First, Observers can explicitly specify which aperture (slit) they wish to use for the TARGET=WAVE exposure (whereas for automatic wavecals the program slit or a pre-defined alternative for each grating is used). Second, they can take longer exposures, increasing the signal-to-noise of the lamp exposures or possibly saturating some lines to bring out weaker lines near astronomical lines of interest. While GO wavecals with the default exposure time are now supported, increasing the exposure time is an available-but-unsupported option (thus requiring justification and approval). The STIS team has developed recommendations for increasing the wavecal exposure times for some of the shortest wavelength settings (see August 2023 STAN).

TARGET=WAVE exposures cannot be taken with all slit-grating combinations. In particular, the line lamps can be too bright for the MAMA detectors when used with wide slits. Therefore only certain aperture-grating combinations can be used for TARGET=WAVE observations. The complete list of allowed combinations is provided on the GO Wavecal Aperture Selection web page. When entering a GO Wavecal in APT, only the permitted slits will be displayed in the APERTURE menu.

Although the slit wheel repeatability can be guaranteed with a high level of precision (see Section Slit and Grating Wheels), observers wishing particularly accurate wavelength calibrations may wish to use a slit for their scientific exposures for which there is an allowed slit-grating wavecal; otherwise, the slit wheel will be moved each time they take a wavecal exposure, producing a small additional uncertainty.

11.2.2 CR-SPLIT

In order to allow rejection of cosmic rays in post-observation data processing, observers using the STIS CCD should always try (as much as possible given signal-to-noise ratio constraints when in the read-noise-limited regime) to obtain at least two—preferably three or more—identical CCD exposures (see Section 7.3.4). In Phase II, the CR-SPLIT optional parameter (default value 2) allows easy scheduling of such multiple associated exposures. You specify the total exposure time and set CR-SPLIT=n, where n is the number of exposures to break the total observing time into. For example, if the total exposure time is 12 minutes and CR-SPLIT=3, then three 4-minute exposures will be taken. Those three exposures will be associated with one another, passed through the STScI calibration pipeline as a unit, and a cosmic-ray–free image will be produced during pipeline processing (see the "STIS Calibration" Chapter of the STIS Data Handbook). Allowed values of CR-SPLIT are integers from 1 to 8. Note that overheads are incurred for each CR-SPLIT subexposure.

11.2.3 Fringe Flat Fields

The STIS CCD exhibits fringing in the far red, limiting the signal-to-noise achievable at wavelengths longward of ~7500 Å in the G750L and G750M spectral modes. As discussed in Section 7.2.6, the best way of eliminating the fringes in the far red is by obtaining contemporaneous flat fields along with the scientific observations. These "fringe flats" must be taken at the same position of the Mode Selection Mechanism as the scientific data. STIS users can insert such contemporaneous fringe flat fields into the same visits as their scientific data, as described below.

Designing Fringe Flat-Field Observations

Observers of extended sources will typically want to take their fringe flat fields using the same slits as they use for their scientific targets, since the flat-field lamp will then illuminate the detector in the most similar way to the targets. However, observers of point sources will typically fare better if they use small slits (e.g., those which are otherwise used for echelle observations) for their fringe flat fields. The main reason for this difference is that the PSF of the STIS CCD features a substantial halo in the far red containing up to 20% of the total source flux, which causes the fringes in lamp flat fields to behave differently from those of external sources, especially in the case of point sources (see also Section 7.2.8). Fringe flat fields taken with short slits simulate the spatial structure of point sources significantly better than those taken with long slits.

The slits supported for scientific observations with the G750L and G750M gratings and the associated slits to use for fringe flat fields in the cases of both extended- and point-source observations in the far red are listed in Table 11.1.

Table 11.1: Slits for Extended-Source and Point Source Fringe Flat Fields.

¹ Short slits are chosen so as to be concentric with the matched long slit. E2 positions are chosen to be concentric with the 52X0.1 aperture at row 900.

A few notes are of importance on the use of short slits for obtaining fringe flat fields:

• Fringe removal for sources that are offset from the center of the long slit will not be possible with a short-slit fringe flat field; one has to use long-slit fringe flat fields for those cases. A special case in this respect is that of point-source spectra with the 52X0.2F1 slit, as the 0.3X0.09 slit (which is in principle the appropriate one to use for fringe flats in that case, cf. Table 11.1) is only a few CCD pixels larger than the occulting bar of the 52X0.2F1 slit. However, a short-slit fringe flat does give a somewhat better fringe correction for the area covered by both the short slit and the 52X0.2F1 slit, so if that area is of particular scientific interest, we recommend taking a short-slit fringe flat as well.
• The limited length of the short slits used for obtaining contemporaneous flat fields of point sources (0.2–0.3 arcseconds) does not allow one to sample the full PSF, so that absolute spectrophotometry cannot be performed with the short-slit fringe flat fields alone. However, a comparison with the pipeline-reduced point-source spectrograms will enable a proper flux calibration.
• At wavelengths longward of ~7500, fringing is the dominant calibration concern at high S/N, whereas imperfect charge transfer efficiency (CTE) is the dominant concern at low S/N ratios. We therefore recommend using the E1 or E2 pseudo-apertures for faint sources and the normal aperture positions in the long slits for high S/N observations.
• The E2 aperture positions are, like the E1 aperture positions, located near row 900 of the detector, and are intended to be used to mitigate CTE effects. However, in order to better align with the 52X0.1 aperture, which is used for fringe flats near row 900, the targeted position is offset about 1 pixel in the dispersion direction from the physical center of each aperture. Fringe-flat alignment will be slightly better than when using the E1 positions, although for the 52X0.2E2 aperture, the throughput will be slightly reduced. The E2 positions should only be used for point source observations where fringe flats are needed and CTE is a concern. If a peakup is desired before using the E2 aperture positions, it should be done using the 52X0.1E1 aperture.
• The limited length of the short slits used for obtaining contemporaneous flat fields of point sources (0.2–0.3 arcseconds) imposes a minimum requirement on the accuracy of the acquisition of target point sources in the slit. The final accuracy should be of the order of 1 pixel (i.e., ~0.05 arcseconds). In case the observer has to use offset acquisition targets, it is therefore recommended that an ACQ/PEAK exposure in a short slit be performed to ensure centering in both directions (see Section 8.3).

Inserting Fringe Flat-Field Exposures in Phase II

A fringe flat-field exposure is specified in the Phase II proposal input as follows:

• Specify Target_Name = CCDFLAT to indicate the exposure as a fringe flat field. The flat-field exposure will automatically be taken at ­CCDGAIN=4.
• Specify Number_Of_Iterations = 2 (to allow cosmic-ray rejection and to obtain adequate signal-to-noise).
• Specify Config, Opmode, Aperture, Sp_Element, and Wavelength.
  • Config must be STIS/CCD.
  • Opmode must be ACCUM.
  • Aperture must be one of 52X2, 52X0.5, 52X0.2, 52X0.2F1, 52X0.1, 52X0.05, 0.3X0.09, or 0.2X0.06.
  • Sp_Element and Wavelength must be one of the following combinations:
    Sp_Element: G750L and Wavelength: 7751
    Sp_Element: G750M and Wavelength: one of 6768, 7283, 7795, 8311, 8561, 8825, 9286, 9336, 9806, or 9851.
• Specify Time_Per_Exposure as DEF (Default). The default exposure time is determined from in-flight calibration data and ensures a signal-to-noise of 100 to 1 per pixel for all settings mentioned above and Number_Of_Iterations = 2.  For very high-S/N observations (e.g., using spatial scans), it may be useful to obtain fringe flats with longer exposure times; see Section 12.12.3 for more details.
  
• If the scientific data are taken in binned mode, specify Optional Parameters BINAXIS1 and BINAXIS2 in the same way as for scientific observations. Supported binning factors are 1, 2, and 4.

Two very important issues for fringe flat fields:

• Fringe flat-field exposures are moved into the occulted period whenever they occur as the first or last exposure in an orbit. Thus the unocculted portion of the orbit can be filled with scientific observations and the fringe flat can be taken during the occultation by placing it at the beginning or end of the orbit.
• Fringe flat fields are effective only if taken without a move of the Mode Selection Mechanism between the scientific exposure and the fringe flat field. Observers must ensure that if the spectral element or wavelength setting is changed during an orbit in which they wish to obtain a fringe flat. They then place the fringe flat-field exposure immediately before or after the scientific exposure(s) they wish to de-fringe. In some cases (e.g., for a long series of exposures), the observer may choose to bracket the scientific exposures with fringe flat-field exposures to be able to account for any thermal drifts.

Please refer to the July 2021 STAN article on defringing and choosing appropriate fringe flats; STIS ISR 1997-15 for more details about NIR fringe flats; STIS ISR 1997-16 for fringing in spectrograms of extended sources; and STIS ISR 1998-19 (Revision A) for fringing in spectrograms of point sources as well as more general fringing analysis and details related to the 52X0.2F1 aperture.  The stistools defringe documentation contains a full overview and user guide for the current defringing tool suite, recently ported from the original IRAF defringe tools into python.

11.2.4 Repeat Exposures

A series of multiple repeated identical exposures can be taken most easily using the Number_Of_Iterations optional parameter in Phase II. In this way, time-resolved observations at minimum time intervals of roughly 20 seconds for the CCD (if subarrays are used) and 30 seconds for the MAMA can be taken in ACCUM operating mode. The output of this mode is a series of identical exposures. For instance, if the exposure time is 60 seconds, and  set Number_Of_Iterations=20, twenty 60-second exposures will be obtained. These twenty exposures will be associated with one another and processed through the pipeline as a unit—the individual exposures will be fully calibrated and a summed image will also be produced for MAMA data as well as a cosmic-ray rejected image for CCD data (see also Chapter 15).