The first line of evidence is the result of a quantitative comparison between the observed morphologies of local galaxies and the predicted appearance of their non-evolved high-redshift counterparts. In order to undertake this comparison we have developed a technique for artificially redshifting local galaxy CCD images by assigning separate spectral energy distributions to individual pixels. Spectral energy distributions for each pixel are determined by using optical colors to interpolate between template spectra corresponding to local S0, Sab, Sbc, Scd, Sdm, and starbursting galaxies. Figure 1 illustrates the power of this technique by showing the excellent agreement between "predicted" and observed far-UV morphologies for a typical galaxy in our calibration sample (1). This figure illustrates the extreme case in which the I-band morphology is extrapolated to the far-UV, corresponding to redshifts z > 4. In fact, at z > 4 the galaxian internal dynamical timescale is a substantial fraction of the age of the Universe, so the existence of morphologically unevolved galaxies is not expected.
Iband image | Bband image |
PREDICTED Far UV | Far UV(UIT) |
Figure 1. Comparison between the far-UV morphology "predicted" on the basis of optical colors of NGC1365 and the observed morphology from the Ultraviolet Imaging Telescope (UIT). Optical images were obtained from the 2.5m du Pont telescope on Las Campanas, and no noise has been added in order to match the signal-to-noise characteristics of the UIT. |
Figure 2. The number-magnitude relations for morphologically segregated samples of galaxies from the HDF and MDS (from Abraham et al. 1996a). Open circles indicate counts obtained from automated classifications, closed circles indicate the results from the visual classifications of Ellis, and crosses indicate the results from the visual classifications of van den Bergh. The MDS counts are indicated by the stars on each panel. The no-evolution = 1 curves from Glazebrook et al. (1995), extrapolated to I = 25 mag, are superposed. The dashed line on the E/S0 diagram shows the effect of assuming = 0.1. The dotted line in panel (a) shows the I-band number counts determined by Smail et al. (1995) from two deep fields imaged with the Keck telescope. |
With this capability to model the effects of bandshifting on observed
morphology, and by assuming an approximate redshift
distribution, one is able account for the impact of morphological
K-corrections on morphologically segregated deep number-magnitude
counts. Because the subjective nature of visual classification makes
comparisons between different groups susceptible to large systematic
errors [20],
the best approach to making these counts is
to adopt a quantitative morphological classification system. Several
quantitative classification systems have now been
developed [1,
3,
21].
Figure 2 shows the number-magnitude counts which
result
from applying a particularly simple system (based on measurements of
central concentration, C, and asymmetry, A) to data from the
Hubble Deep Field. Also shown are the no-evolution predictions for
ellipticals, spirals, and irregular/peculiar/merger systems,
constructed as described in
Glazebrook et
al. (1995) and
Abraham et al. (1996a,
b),
by adopting Schechter luminosity
functions (LFs) with parameters given by
Loveday et al. (1992),
and a high normalization = 0.03 h3
Mpc-3. The
predicted counts for the elliptical galaxies are based on a flat slope
( = -1) for the faint-end of
the LF, rather than the turn-over
originally found by Loveday et al. The steep counts for the
irregular/peculiar/merger systems continues to the limits of the
survey. Beyond I814 = 22 mag the spiral counts show a
significant excess over the no-evolution predictions. A weaker trend
is seen for the spheroidal systems (whose counts are only marginally
above the no-evolution prediction) and there is some evidence of a
turn-over in the last magnitude interval.
1 In order to make a
comparison with UIT straightforward, noise has not been added to the
simulations. Because of (1 + z)4 cosmological dimming
extremely long
exposure times with HST would be required to image this galaxy at high
redshifts with a signal-to-noise level equivalent to that
shown. Back.