Figure 2.5. An example of dispersed images obtained with the HST and NICMOS during a blind slitless spectroscopic grism survey for Ha emitters at high redshift. Both undispersed (left) and dispersed (right) images are shown, as well as the location of the emission lines and the order of the spectra. From McCarthy et al. (1999).
imaging surveys, because the dispersed images occupy more detector area, and because confusion from adjacent overlapping spectra makes some portion of the dispersed image useless. Observations are generally easier than slit spectroscopy, but data reduction and especially analysis (calibration and extraction of the spectra) are usually complex. Figure 2.5 shows, as an example, the grism images taken as part of the Ha survey by McCarthy et al. (1999) with the Hubble Space Telescope (HST) and NICMOS. Both undispersed (left) and dispersed images (right) are shown. Figure 2.6 shows a gallery of grism spectra to illustrate the type of results and the diversity of spectra encountered in such surveys, while Figure 2.7 shows examples of extracted spectra.
2.3.3 The sky background, Finding distant galaxies is a contrast problem. The night sky, especially at the near-IR wavelengths to which the target emission lines are redshifted, is much brighter than the galaxies. Figure 2.8 shows the night-sky spectrum in the optical window, together with the wavelength ranges of traditional filters and the wavelengths of important emission lines, while Figure 2.9 shows a synthetic spectrum of the night sky in the near-IR windows (from Rousselot et al. 2000) with the relative intensity of the lines proportional to the photon flux.
Since, as the figures show, the atmospheric OH emission lines generally dominate this background in broad-band filters, as well as in narrow-band ones (unless they are very narrow and/or strategically positioned, as we are about to discuss), one way to make ground-based observations more sensitive is by adopting transmittance curves limited to wavelength intervals that fall between the molecular-line complexes. For example, narrow-band imaging through one of the larger OH windows at 8200 A led to the first samples of Lya-selected galaxies at redshift 5.7 (Hu et al. 2004; Rhoads et al. 2003; Taniguchi et al. 2003). Figure 2.1 illustrates this strategy in the implementation by Hu et al. (2004).
In principle, the sensitivity of observations over a relatively small wavelength interval can also be improved by dispersing the light. For example, if a final resolution of, say,
3 A is adequate for the emission lines targeted by the observations, the 150-A-wide, OH-free window around 8200 A discussed above can be dispersed with a spectrograph to improve the line-to-sky contrast by reducing the sky brightness by another factor of 50 (i.e. 150 A/3 A FWHM). Custom-designed, multi-slit masks can be employed in this case, when spectroscopic observations are made with a band-limiting, OH-suppression filter. This multi-slit-windows technique had been used previously to search for z = 5.7 and z = 6.5 Lya emitters using 8-meter-class telescopes (Crampton 1999; Stockton 1999; Martin & Sawicki 2004; Tran et al. 2004). However, although some high-redshift Lya emitters have been discovered serendipitously during deep blind spectroscopic searches, the volume probed in a typical long-slit observation is very small. The authors of the surveys mentioned above did not report detections of Lya emitters, very likely because they did not probe large-enough volumes of space.
Was this article helpful?