16.1 Summary of Accuracies

In this chapter we summarize the typical accuracies achieved by the STIS photometric, spectral, and astrometric calibrations. Tables 16.1 through 16.5 list the expected accuracies for each of the basic STIS observation modes: CCD spectroscopy, MAMA spectroscopy, CCD imaging, MAMA imaging, and target acquisition. All the accuracies correspond to 2-sigma limits; for the MAMA detectors, the pixels are the "native" low-resolution pixels. The specific values given in the tables are those characterizing the "close-out" calibration of data obtained prior to the suspension of STIS operations early in Cycle 13, reflecting our understanding of STIS as of 2010 August. The listed values thus are those to be expected for the pre-SM4 data that were delivered to the archive under that close-out calibration. Following the repair of STIS in SM4, both the CCD and the MAMA detectors have exhibited somewhat increased dark rates, (continued) slow declines in sensitivity, and (for the CCD) reduced charge transfer efficiency (Sections 7.2 through 7.4). After accounting for those effects, however, similar accuracies generally have also been achieved for post-SM4 STIS data. The sources of inaccuracy are described in Chapter 4 of the STIS Data Handbook, which includes discussions of various instrumental phenomena and the creation of reference files that characterize those phenomena. More detailed discussions may be found in various STIS Instrument Science Reports (ISRs) and STScI Analysis Newsletters (STANs).

The absolute and relative flux accuracies quoted in Tables 16.1 and 16.2 apply only to spectroscopic observations using relatively large apertures (52 × 2" for 1st order modes; 0.2 × 0.2" for the echelle modes).  For smaller apertures, however, the reliability and repeatability of the throughput has been less well quantified. There is some evidence that the average STIS focus, relative to that of other HST instruments, changed between 2013 and 2015 (STIS ISR 2017-01). For apertures less than or equal to about 0.1" in size, that change in focus appears to have resulted in both decreased average throughput and increased throughput variability. Analysis of echelle observations taken through the 0.2 × 0.06" and 0.2 × 0.09" apertures indicated that the average throughput was only about 80% of nominal, with some individual exposures showing as much as a 40% throughput loss; for the smallest aperture (0.1 × 0.03"), the average throughput was typically only half its nominal value. (Note that for these apertures, throughput variations of order 10% due to telescope breathing have been commonly observed throughout the lifetime of STIS.)  After 2016, however, the average STIS focus appears to have recovered somewhat (STIS ISR 2019-01).  Because the throughput losses can vary significantly from observation to observation, it is not possible to simply update the ETC throughputs, as the ETC must also warn against observations which are too bright or which may cause saturation, and must therefore adopt the highest throughput that might reasonably be obtained. Focus offsets can also affect the relative flux calibration as a function of wavelength within a given observation. For modes covering a wide range of wavelengths, relative flux errors of 10% over the wavelength span of E140M and E230M observations done with the 0.2 × 0.06" aperture are not uncommon. If combined with small aperture centering errors, the relative throughput errors can sometimes increase to as much as 25%. The most up-to-date information about STIS focus can be found at HST's Focus page, which includes a link to an interactive focus model. 

Many significant changes in pipeline calibration have been made during the lifetime of STIS—reflecting both better understanding of the instrument and compensation for observed changes in its behavior; see Chapter 3 of the STIS Data Handbook for details. Extracted spectra and rectified spectral images from all three STIS detectors are now corrected for both time-dependent and temperature-dependent variations in sensitivity (STIS Data Handbook Section 3.4.13; STIS ISR 2017-06). Extracted CCD spectra are corrected for CTE losses (STIS ISR 2022_07) and are adjusted for the formerly neglected interdependence of grating and aperture throughputs. Time-dependent rotation of the spectral traces is applied to the most commonly used first order modes during spectral extraction and spectral image rectification (STIS ISR 2007-03). Both the echelle blaze shift correction (for spectral extractions) and the echelle flux calibration have recently been substantially improved (e.g., 2017 August STAN; STIS ISR 2022-04). The flat-field reference files have also been revised (STIS Data Handbook Section 4.1.4).  Recent improvements to the stellar atmospheric models for the primary flux standard stars, incorporated into the CALSPECv11 database, have prompted an effort to update the flux calibration and zero points for all of the STIS spectroscopic and imaging modes.  The revised calibrations for the highest priority spectroscopic modes were delivered in 2022 April, and work is continuing on the rest of the modes (see https://www.stsci.edu/hst/instrumentation/stis/flux-recalibration).

We remind users that calibration data have always been immediately non-proprietary. If users have a need for higher accuracy or urgent results, they may wish to consider direct, custom analysis of the calibration data for their particular observing mode. See also Chapter 17 of this handbook for a description of our regular on-orbit calibration program.

Table 16.1: CCD Spectroscopic Accuracies

¹ For more recent analyses of wavelength accuracy, see STIS ISR 2011-01, STIS ISR 2015-02, and STIS ISR 2018-04. Note that the wavelength accuracies will also depend on the accuracies of the rest wavelengths used in calculating the dispersion relations.  The quoted accuracies refer to spectra obtained at the nominal (central) locations; the wavelength zero points may be somewhat less accurate for recent spectra obtained at the E1 pseudo-aperture.
² Assumes star is well centered in slit and use of a 2 arcseconds wide photometric slit. See the STIS Data Handbook for a more complete description of the impact of centering and slit width on accuracies. This accuracy excludes the G230LB and G230MB modes when used with red targets, for which grating scatter can cause large inaccuracies in the flux calibration at the shortest wavelengths; see Gregg et al. 2006 (HST Calibration Workshop) and STIS ISR 2022-05. Photometric accuracies referenced are for continuum sources; equivalent width and line profile measures are subject to other uncertainties (such as spectral purity and background subtraction).

Table 16.2: MAMA Spectroscopic Accuracies

¹ For more recent analyses of wavelength accuracy, see STIS ISR 2011-01, STIS ISR 2015-02, STIS ISR 2018-04, and Ayres 2022. Note that the wavelength accuracies will also depend on the accuracies of the rest wavelengths used in calculating the dispersion relations.
² A pixel for the MAMA refers to 1024 × 1024 native format pixels.
³ Assumes star is well centered in slit and use of a wide photometric slit.
⁴ For 0.2 × 0.2 arcsecond slit. These are typical accuracies (which can be 2 to 3 times better or worse as a function of wavelength; see STIS ISR 1998-18 for details).

⁵ Quoted relative flux accuracies of echelle spectra assume that the time dependent shifts in the echelle blaze function are properly corrected. Recent improvements to the blaze shift correction yield agreement in the order overlap regions to better than 5% for E140H (see August 2017 STAN).

Table 16.3: CCD Imaging Accuracies

Table 16.4: MAMA Imaging Accuracies

¹ A pixel for the MAMA refers to 1024 × 1024 native format pixels.
² A recent re-analysis of the FUV-MAMA geometric distortion has yielded rms residuals of 4mas (0.16 pix) in each coordinate, compared to the positions in an astrometric standard catalog based on WFC3/UVIS imaging data see (August 2017 STAN and STIS ISR 2018-02).

Table 16.5: Target Acquisition Accuracies