CLASSIFICATION OF EXTREMELY RED OBJECTS IN THE COSMOS FIELD

We present a study of the classification of z similar to 1 extremely red objects (EROs), using a combination of Hubble Space Telescope (HST) Advanced Camera for Surveys (ACS), Spitzer Infrared Array Camera (IRAC), and ground-based images of the COSMOS field. Our sample includes similar to 5300 EROs with i - K-s >= 2.45 (AB, equivalently I - K-s = 4 in Vega) and K-s <= 21.1 (AB). For EROs in our sample, we compute, using the ACS F814W images, their concentration, asymmetry, as well as their Gini coefficient and the second moment of the brightest 20% of their light. Using those morphology parameters and the Spitzer IRAC [3.6]-[8.0] color, the spectral energy distribution (SED) fitting method, we classify EROs into two classes: old galaxies (OGs) and young, dusty starburst galaxies (DGs). We found that the fraction of OGs and DGs in our sample is similar, about 48% of EROs in our sample are OGs, and 52% of them are DGs. To reduce the redundancy of these three different classification methods, we performed a principal component analysis on the measurements of EROs, and find that morphology parameters and SEDs are efficient in segregating OGs and DGs. The [3.6]-[8.0] color, which depends on reddening, redshift, and photometric accuracy, is difficult to separate EROs around the discriminating line between starburst and elliptical. We investigate the dependence of the fraction of EROs on their observational properties, and the results suggest that DGs become increasingly important at fainter magnitudes, redder colors, and higher redshifts. The clustering of the entire EROs, DGs, and OGs was estimated by calculating their correlation function, and we find that the clustering of EROs is much stronger than that of full K-limited samples of galaxies; the clustering amplitude of OGs is a factor of similar to 2 larger than that of DGs.