Spectroscopy must be planned with several factors in mind. Taken together, these factors require one to make trade-offs in selecting the number and duration of exposures, the placement of exposures on the detector, and the observing configuration. In some cases, extra calibration exposures may be needed in addition to the calibration reference files provided by STScI.
Spectroscopy Strategies with CCD
Read noise: Read noise can be significant in observations of faint objects. Fortunately, read noise is less for CCDGAIN=1 than for CCDGAIN=4. On-chip binning has been recommended to reduce the read noise by the binning factors. Binning, however, also increases the percentage of the image affected by hot pixels (which are becoming more numerous) and cosmic rays. Binning thus becomes less desirable with later epoch and longer exposure time. The problem is further exacerbated when bilinear intepolation spreads defects during rectification of a spectral image.
Hot pixels: The number of hot pixels has been steadily increasing, making on-chip binning less favorable and dithering more desirable. If exposures are dithered by a few pixels, hot pixels can be rejected like cosmic rays when the images are registered and combined. Dithering can be implemented in Phase 2 with the use of patterns.
Cosmic rays: Two or more exposures should be taken, either with CR-SPLIT or in combination with dithering, so that cosmic rays can be rejected when the images are combined. Dithering can be implemented in Phase 2 with the use of patterns.
Charge transfer efficiency: CTE of the CCD detector steadily decreases with time in orbit due to radiation damage. Charge transfer is more efficient for greater count levels in the target and in the background, so longer exposure times are helpful. Losses are less for sources in rows closer to the readout amplifier. Since Cycle 10, it is possible to use aperture names for the STIS long slits which place the target high on the detector, close to the default amplifier, to reduce loss.
Saturation: Saturation occurs at a higher level for CCDGAIN=4 than for CCDGAIN=1. Nonlinear amplification occurs at levels somewhat below saturation for CCDGAIN=1. Saturation does not damage the detector, so it can be tolerated if it does not occur at locations of interest or bleed charge into them along columns.
PSF sampling: The CCD marginally undersamples the HST PSF at optical wavelengths. As a result of this undersampling, interpolation on the spatial axis during rectification of a spectral image produces artifacts in the final image of a target that includes an unresolved component. Rectification can be improved by subpixel dithering along the spatial axis. Dithering can be implemented in Phase 2 with the use of patterns.
Effects of slit width: It may be advisable to choose a wider slit to obtain better signal-to-noise or greater photometric accuracy. Before doing this, however, one should consider the impacts of degraded line profiles (wings added for point sources, degraded resolution for extended sources), contamination by scattered light, and the degeneracy of wavelength differences and spatial offsets across the slit.
Flat field accuracy: Reference file flat fields are not adequate to remove fringing from spectral images taken with G750L or G750M for wavelengths longward of about 7500 Å; observers should take contemporaneous fringe flats. Guidance for observations and reductions are given in ISRs 1997-15, 1997-16, 1998-19, and 1998-29. Even at shorter wavelengths, flat fielding can be the dominant source of error in observations with very high signal-to-noise. Dithering can improve the signal-to-noise by smoothing out flat field errors.
Wavelength accuracy: Automatic wavecals are sufficient for most purposes, but some observers with special wavelength calibration needs may want to add wavecals or replace the automatic wavecals to control the timing, duration, and aperture selection.
Escaping MAMA bright object limits: In the case of near UV observations, if the target or a nearby object violates the MAMA bright object limits, you may choose to switch from the NUV-MAMA with G230L or G230M to the CCD with G230LB or G230MB.
Spectroscopy Strategies with MAMA
Bright object limits: Local and global count rate limits must be met for the aperture area at any position angle allowed for the exposure, and within 5 arcsec beyond that area. Less severe limits must be met within an additional buffer zone. There are several strategies for reducing the flux received from a bright target or bright companion object. Additionally, you may be able to exclude a bright companion object from the critial area by selecting a limited position angle range (ORIENT range) for the aperture. Application of MAMA bright object limits to solar system observations can be found in the STIS Instrument Handbook.
Time resolution: Time-resolved observations can be made either in ACCUM mode or in TIME-TAG mode. The minimum sampling time is shorter in TIME-TAG mode, but other timing constraints can also affect the choice of mode . For TIME-TAG mode, one must choose the BUFFER-TIME to control the frequency of data transfer. TIME-TAG analysis is described in STIS ISR 2000-02.
Saturation: The STIS internal buffer limits MAMA images to 65,536 photons per pixel.
PSF sampling: PSF sampling can be improved by using the highres mode image, at the cost of flat field accuracy and signal-to-noise. Native format pixels marginally undersample the HST PSF. As a result of this undersampling, interpolation on the spatial axis during rectification of a spectral image produces artifacts in the final image of a target that includes an unresolved component. Rectification can be improved by subpixel dithering along the spatial axis. Dithering can be implemented in Phase 2 with the use of patterns.
Effects of slit width: It may be advisable to choose a wider slit to obtain better signal-to-noise or greater photometric accuracy. Before doing this, however, one should consider the impacts of degraded line profiles (wings added for point sources, degraded resolution for extended sources), contamination by scattered light, and the degeneracy of wavelength differences and spatial offsets across the slit.
Flat field accuracy: Flat fielding can be the dominant source of error in high signal-to-noise observations. Dithering can improve the signal-to-noise by smoothing out flat field errors. For echelle observations, dithering can be achieved by observing with fixed-pattern slits (FP-SPLITs).
Wavelength accuracy: Automatic wavecals are sufficient for most purposes, but some observers with special wavelength calibration needs may want to add wavecals or replace the automatic wavecals to control the timing, duration, and aperture selection.