Astron. Astrophys. 326, 950-962 (1997)

4. Calibration of bright counts

Because they do not depend on the cosmological parameters, bright galaxy counts () are a straightforward constraint on spectral evolution models and notably on the reliability of the templates. Two related problems concerning bright counts arose in previous analyses (Guiderdoni & Rocca-Volmerange 1990; Pozetti et al. 1996), namely the slope at bright magnitudes and the normalization of the luminosity function. Maddox et al. (1990) have found in the APM survey a steeper slope of number counts between and than allowed by pure luminosity evolution models, implying a much more rapid evolution at low redshift. As a result, when model predictions are calibrated on observed counts in the brightest bins, a strong deficit of predicted counts relative to observed ones is obtained at faint blue magnitudes. Although the corresponding low normalization of the luminosity function is in agreement with Loveday et al. (1992), most models have preferred until now to normalize the counts at fainter magnitudes (), where evolution may become non-negligible, to recover agreement with faint counts.

This is clearly a crucial problem for the reliability of evolution models. However, recent observational results do not confirm the conclusions of Maddox et al. (1990). The re-analysis of the photometry of APM data by Metcalfe et al. (1995) and the counts of Bertin & Dennefeld (1996) favor a flatter slope and a high normalization of the luminosity function. To clarify this controversy, a multispectral analysis of bright counts is needed, highlighting by different weights of old and young stellar populations in the various bands, whether a strong evolution has happened recently or not. The NIR, in particular, is dominated by slowly-evolving old stellar populations and should therefore be more sensitive to number density evolution than to recent star formation. Moreover, in this wavelength range, the k-corrections for different types are very similar, contrary to the UV. Finally, it is less affected than the blue by the uncertainties in the luminosity functions of late-type galaxies. For these reasons, predictions in the K -band are more reliable than at shorter wavelengths. The evolutionary spectra built by PEGASE are particularly useful for this purpose, thanks to their large continuous range of wavelength and valid fits of nearby observational templates.

We have modelled galaxy counts in the 3 bands , I and K (Fig. 9). The adopted luminosity function is that of Marzke et al. (1994), after correction of the Zwicky magnitudes by , the mean value obtained by Efstathiou et al. (1988), to convert them in B magnitudes. Characteristics of the luminosity functions are given in Table 3. When necessary, we redistribute the types given for the LF in our standard types. We take in agreement with Tanvir et al. (1995). An open cosmology (, ) is assumed, leading to an age of the universe of 13.5 Gyr greater than the age of our older standard galaxies, but does not in any way affect the simulation of bright counts. Redshifts of formation in this cosmology (see Table 3) are taken in agreement, within the uncertainties, with the ages of the reference model spectra. We finally normalize our counts on Bertin & Dennefeld (1996) counts at .

Fig. 9. Galaxy counts in , I and K with the Marzke et al. (1994) luminosity function. Cosmological parameters are , and . The solid (resp. dashed) lines are our predicted counts with (resp. without) evolution.

Table 3. Redshifts of formation and luminosity functions (Schechter parametrization) of our standard synthetic spectra. is given for . Two LFs are given for Sa and should be added. The reason for this is that the S0 type of Marzke et al. (1994) includes S0/a, while these are included in our synthetic Sa. Part of the S0 LF is for this reason transferred to the synthetic Sa.

We obtain good agreement with Gardner et al. (1996) multispectral bright counts in the 3 bands , I and K for the same value of normalization, confirming that the evolution scenarios and colors of the galaxies dominating bright counts are correct. Moreover, the value of from modeling is fully consistent with the normalization of the Marzke et al. (1994) luminosity function that we use.

Predictions at fainter magnitudes in various cosmologies and constraints from redshift and color distributions as well as correlation functions will be extensively discussed in a future paper.

© European Southern Observatory (ESO) 1997

Online publication: April 8, 1998
helpdesk.link@springer.de