Making use of the Bayesian technique, Benítez (2000) demonstrated that the dispersion of can be significantly improved. Despite of this result, we decide not to introduce this possibility in our code, at least for general purposes. The reason for this is that we want to prevent spurious effects in particular studies. As an example, when the luminosity function is imposed, the study of the galaxy population is constrained and it becomes impossible to obtain independent information on the properties of objects, thus limiting the possible applications. However, this method can be regarded with interest when the purpose is addressed to some specific application or when one is dealing with poor data, in such a way that the introduction of hints permits to obtain useful results. Alternatively, the photometric redshift estimate can be safely improved introducing the Bayesian inference when prior information is not related to photometric properties of sources. Examples of such priors that could be combined with the technique are the morphology or the clues inferred from gravitational lensing modeling.
One of the main issues for is the optimization of the visible versus near-IR bands for spectroscopic surveys. The aim is to produce a criterion based in to discriminate between objects showing strong spectral features in the optical and in the near-IR. To perform this test, both the redshift and the SED characteristics have to be estimated for each object. The and the SED are obtained by means of hyperz , together with the best fit parameters (, spectral type, metallicity and age). The relevant information shall be the redshift and the rough SED type, i.e. "blue" or "red" continuum at the given z. We have shown that only limited information could be obtained on the parameter space from broad-band photometry alone. This situation will change with the future cryogenic imaging spectrophotometers, as presented in a recent paper by Mazin & Brunner (2000), because such devices will be able to gain in spectral resolution while spanning a large wavelength domain.
Another important issue for is the improvement on the cluster detection in wide-field photometric surveys. Including such a technique in an automated identification algorithm, whatever this algorithm is, allows to improve significantly the detection levels. The main idea is that the contrast between the cluster and the foreground and background population is the leading factor. When introducing a simple detection scheme, similar to the one used by Cappi et al. (1989), it is easy to quantify this effect (Pelló et al. 1998). In general, the is expected to improve by a factor of at least to 3 with respect to the pure 2D case, depending on the cluster redshift and richness, the set of filters used and the depth of the survey. When considering more elaborated cluster-finding algorithms, such as the one produced by Kepner et al. (1999), Olsen et al. (1999), Scodeggio et al. (1999), Kawasaki et al. (1998) or Deltorn et al. (2000, in preparation), these results could be regarded as the relative improvement due to photometric redshifts. The present version of hyperz is also able to display the probability of each object to be at a fixed redshift. This is useful when looking for clusters of galaxies at a given (guessed) redshift.
The study of clustering properties through the spatial correlation function of galaxies, using the angular correlation together with the information is another possible application of , aiming to extend the study of galaxy properties to fainter limits in magnitude. In this case, the relatively high number of objects accessible to photometry per redshift bin, suitably defined according to photometric redshift accuracy, allows to enlarge the spectroscopic sample towards the faintest magnitudes, and also to strongly reduce the errors (because the number of objects per redshift bin strongly increases). Studies on the evolution of the angular correlation function of galaxies in the HDF-N applying the photometric redshift technique can be found in Miralles & Pelló (1998), Connolly et al. (1998), Roukema et al. (1999), Arnouts et al. (1999), Magliocchetti & Maddox (1999).
The same slicing procedure can be adopted to study the evolution of the luminosity function and consequently to infer the star formation history at high redshift from the UV luminosity density, as well as to analyse the stellar population and the evolutionary properties of distant galaxies (e.g. Yee et al. 1996; SubbaRao et al. 1996; Gwyn & Hartwick 1996; Sawicki et al. 1997; Connolly et al. 1997; Pascarelle et al. 1998; Giallongo et al. 1998).
Furthermore, the photometric redshift method has been used to investigate the nature of Extremely Red Objects (EROs) with a "spectro-photometric" technique by Cimatti et al. (2000), deducing clues about the model of galaxy formation. Another kind of spectroscopic and photometric combination has led to the identification of very high redshift object, as described by Chen et al. (1999).
From this not exhaustive list of applications, it is evident that photometric redshifts are a powerful and promising tool in many areas of extragalactic research. This method shall not be regarded only as a "poor person's redshift machine", but as a fundamental instrument, since a multitude of faint objects will remain beyond the limits of spectroscopy for the next years. Even with the diffusion of Multi-Object Spectrometers, most of the faint galaxies with measured photometry will fall beyond the reach of conventional spectroscopy.
© European Southern Observatory (ESO) 2000
Online publication: December 11, 2000