Astron. Astrophys. 341, 641-652 (1999)

5. The multicolor selection for high redshift galaxies

Steidel & Hamilton (1993) used a combination of three filters U, G and to select star forming galaxies at 2.8 3.4. At these redshifts, due to the shift of the Lyman continuum break in the UV band, the U flux is attenuated and the color is strongly reddened while the color still reflects a flat spectrum.

To select galaxies at higher redshifts, a different set of filters is required. At 4, the combination of Lyman clouds absorption and the Lyman continuum reduces significantly the observed flux also in the B filter. Giallongo et al. (1998) have shown that an optimized set of filters can be found to detect star forming galaxies at and to reduce the confusion with low redshift galaxies of similar colors. This set is based on the B,V,r and I bands, and corresponds to the one adopted in the present work. An optimized detection of star forming galaxies in the redshift range 3.84.4 can be obtained by using the following criteria: 2, 0.5 and 0.1. In Fig. 9, we have applied the criteria to our catalogue. Two approaches have been followed to select galaxy candidates up to .

• A: Galaxies with measured colors (or upper limits) located inside the area defined by the above criteria (8 galaxies).

• B: The above sample and galaxies with upper limits falling less than outside the selection criterion (15 galaxies).

Fig. 9. Color criteria to select star forming galaxies at redshift 3.8 4.4. Triangles show objects with upper limits distant less than 1 from the color selection criterion, circles show objects with upper limits distant between 1 and 2 from the selection criterion. Both are considered as upper-limits (defined with arrows). The galaxy with B and V upper-limit and filled triangle (# 596 in the catalogue) is spectroscopically confirmed at . Stars (with stellarity index 0.9) are shown with star symbols. For clarity, error-bars are put only for galaxies matching the color criteria.

The photometric redshifts for the galaxies of this field will be estimated in a forthcoming paper (Fontana et al., 1998) by comparing the colors (including J and K) with a library of synthetic galaxy spectra as described by Giallongo et al. (1998). Due to their faint magnitudes, the spectroscopic confirmation of our candidates requires an 8-m class telescope. Only one of the galaxies (object # 596 in our catalog) is included in the spectroscopic sample by Hu et al. (1998) and Hu (1998,private communication) and has a spectroscopic redshift of . This galaxy has upper limits in the B and V bands and a magnitude .

The number of galaxy candidates in different magnitude intervals are given in Table 3 and Table 4 for both classes A and B described above. The surface density at 25, is in the range 0.7-0.9 arcmin^-2 (case A and B respectively). At 26, the density increases to 1.4-2.7 arcmin^-2.

Table 3. Number distribution of galaxies in redshift range located inside the color selection criteria (case A with 8 objects).

Table 4. Same as Table 3 for case B (with 15 galaxies).

Assuming a uniform redshift distribution of the sample in the range 3.84.4, the comoving galaxy number density at for is estimated to be (q₀=0.5) and (q₀=0.5) for r.

The star formation rate history at 4.1 can be derived from the luminosity in the I band corresponding to the restframe wavelength 1500 Å. We adopt the conversion factor (from the UV luminosity at 1500 Å to the SFR) of Madau et al. (1996). For a Salpeter IMF (0.1) with constant SFR, solar metallicity and age range of 0.1-1 Gyr, a galaxy with produces .

The brightest galaxy in our sample has a magnitude of corresponding to a for (), and the faintest one has a magnitude of corresponding to a for () (the uncertainties are linked to the redshift range ).

Assuming that the redshift range is uniformly probed, the star formation rate per unit comoving volume at derived from objects with varies (for cases A and B, respectively) between with (). For , the estimates increase to with ().

This latter estimate is close to the value of computed by Madau (1997) for the HDF in the same redshift interval down to

© European Southern Observatory (ESO) 1999

Online publication: December 16, 1998
helpdesk.link@springer.de