The signal-to-noise ratio (S/N) of low-dispersion spectra of sufficiently bright targets may be improved by trailing the target along the long axis of the large aperture, perpendicular to the dispersion direction. Such trailed spectra, if optimally exposed, have S/N of nearly two times that of untrailed spectra. The exposure time required for a trailed exposure is 3.7 times that for an optimum point-source exposure.
Trailed exposures may also be performed in high dispersion by trailing the star along the short axis of the large aperture. However, this reduces the width of the interorder background required to reduce these echelle spectra. Since the crowding of the orders is greatest at shorter wavelengths, high dispersion trails should be performed only if the spectral features of interest are at relatively long wavelengths. The improvement in S/N is about 1.4 for high-dispersion spectra which are optimally exposed. The exposure time required for a trailed exposure is 1.85 times that for an optimum point- source high-dispersion exposure.
Equations for computing the trailed exposure times and trailing rates are given below. Point-source exposure times may be estimated as described in Section 3.10, or determined from previous exposures.
low dispersion: t(tr) = 3.7·t(point)
Trail rate = 20.0/t(tr) arc seconds/second
high dispersion: t(tr) = 1.85·t(point)
Trail rate = 10.0/t(tr) arc seconds/second
Trailed exposures have greater observing overhead than point-source spectra. Generally, they require a total of from 10 to 15 minutes plus twice the trailed exposure time to perform. Rates of 0.03 to 25.0 arc seconds/second may be used for normal trails, but those performed at 0.1 to 10 arc seconds/second are more reliable. This corresponds to a trailed exposure time range of from 2 seconds to 3.3 minutes. More than one pass may be used to increase the exposure time. While the trail is being performed on one camera, it is not possible to read or prepare the other camera. In general, the target must be stellar and relatively bright (< 10th magnitude) so that the FES can track it accurately. Despite increased overhead, trailed exposures are more efficient than adding together four exposures read down separately, and result in comparable signal-to-noise. In addition, because the trailed spectrum is spread across a larger portion of the camera faceplate, some sources of camera noise (such as fixed-pattern noise) are averaged out.
For trailed exposure times between 0.2 and 2 seconds, a "fast trail" technique is available to obtain near-optimal exposures for very bright stars, which would otherwise be overexposed at the shortest point-source exposure time of 0.41 seconds (1.52 second trailed exposure). For optimal results, these trails should be performed when the target is at a Beta angle of 90±10 degrees. For an analysis of fluxes obtained with this technique see Oliverson (1986).
For trailed exposures times longer than approximately 10 minutes, multiple exposures may be placed side by side in the large aperture producing a "pseudotrailed" exposure. Usually three or four such spectra are placed within the aperture. The pseudotrail can often be performed with little extra overhead and yields a similar factor of two improvement in S/N. Operationally, a multiple exposure is specified by the "offset reference points," where the target is positioned prior to being placed in the aperture. The observer should specify the offset reference points to be used in the appropriate section of the observing script, a sample of which is shown in Figure 4.5. The exposure time for each segment is the point-source exposure time. The RA can recommend which offset reference points to use for a given type of observation.
Figure 4.5: An observing script for a multiple or "pseudotrailed" exposure. The requested offset reference points are written in the appropriate section. Three point-source spectra of 10 minutes each were obtained for a total exposure time of 30 minutes which is the value listed in the merged log. Positioning of the target in the aperture corresponding to the offset reference points was performed by displacing the target by the appropriate number of fine FES units in Y.
For special purposes, such as time-critical observations, it is often possible to place two separate exposures within the large aperture at low dispersion (e.g. Holm et al. 1985). The star is offset by 5 or 6 arc seconds towards either end of the aperture for the two exposures. The amount of the separation is not sufficient in eliminating overlap of the two spectra. The standard image processing system cannot extract the individual spectra; special user techniques are therefore required to do so. Good results have been obtained by fitting Gaussian profiles to cuts across the spectra (Panek and Holm 1984).
Information for deriving fluxes from trailed spectra is given in Section 5.3.