7. Redshift determinations
A major problem when comparing predictions from different scenarios to observations is the determination of the SN redshift. A direct spectroscopic determination with a resolution of 100 is only possible for SNe more than two magnitudes brighter that the limiting magnitude, i.e. IAB 25 for m class telescopes and for the NGST. To reach fainter magnitudes the main alternative is photometric redshifts of either the host galaxy or the SN. Photometric redshifts for galaxies have been discussed extensively by e.g., Fernández-Soto et al. (1999), Yee (1998), Gwyn (1995) and Connolly et al. (1995). The fact that a large fraction of the star formation up to 1 occurs in dwarf galaxies, as well as the cosmological dimming, can make such a determination difficult. An alternative is to estimate a photometric redshift directly from the SN. A problem here is that the SN spectrum changes with both type and epoch.
To examine this possibility we have determined broad band colors for the different types of SNe as function of epoch and redshift. As an example we show in Fig. 10 the color indices for a Type IIP SN as function of redshift at different phases. The spectra are taken from the synthetic spectra calculated by Eastman et al. (1994). These spectra assume LTE, but may nevertheless give a good impression of the qualitative evolution.
At early time, before 20 days the spectrum is a fairly smooth blackbody without a strong UV cutoff. The color indices do therefore not change dramatically with redshift. At later stages in the plateau phase the spectrum does not change much. An important aspect is that the UV cutoff at 4000 Å has now developed to its full extent, and the UV is essentially black. This is probably the most useful feature for identifying high redshift SNe photometrically. The extent of this UV drop may, as we discuss below, depend on the metallicity. The UV cutoff has a very pronounced effect on the optical color indices at 1, with strong increases in the B-I, V-I and R-I indices at successively larger z. For 1 the J-I and K-I, and finally K-J, are most useful due to the essential disappearance of the SNe in the optical.
A problem with photometric redshifts of SNe, compared to galaxies, is that the colors, as we have seen, depend sensitively on the epoch. In addition, they depend on SN type. E.g., Type IIL's have less UV blanketing, while the Type Ia's have a rapid development of the UV cutoff. To break this strong degeneracy it is essential to have information about both SN type and epoch. It is therefore necessary to obtain reasonable light curves, i.e. a fairly large number, 5-10, observations of the field. A complete analysis can therefore be quite costly in terms of observing time.
An alternative redshift method may be to use reasonably well sampled light curves in combination with the cosmic time dilation. For Type Ia SNe one can safely assume a standard light curve. Although the absolute luminosity can vary by a large factor, Type IIP's have a fairly well defined duration of the plateau phase, which lasts 100 days. Also Type IIL's and Ib's and Ic's have reasonably standardized light curves. From the observed light curve one can then, at least for SNe with 1, get an approximate redshift within 25% from a comparison with the low z light curve templates. However, since the light curve should be followed over a decline of about 2 magnitudes in order to achieve a type specific light curve, the gain in depth by using photometry instead of multi object spectroscopy is marginal. The photometric accuracy also decreases for these levels and the actual limit may be even higher than two magnitudes. Therefore, in practice little is probably gained compared to direct spectroscopy.
© European Southern Observatory (ESO) 1999
Online publication: October 4, 1999