Deep NICMOS/HST and radio 1.4 GHz observations of galaxies at high redshifts

An increasing number of observations have shown that dusty starburst. galaxies are probably much more numerous at high redshifts than today, and optical surveys of the distant universe suffer from large extinction corrections. In this talk, I present the quantitative estimate on how much dust extinction correction needed to apply to the rest-frame UV luminosity density at z similar to 1.3. In addition, I will discuss the recent deep 1.4GHz radio observations of extremely red galaxies (EROs), covering an area of 26 x 26 with complimentary deep optical and near-IR images. We found that the fraction of bright EROs (H G 20) detected in deep 1.4GHz images is small, only 8-17%. The implication is that a large fraction of bright EROs are probably old ellipticals or systems with a small amount of star forming activities at z similar to 1 - 2. We found that similar to 20% of well detected micro-Jansky radio sources are very faint or even not detected in the optical and near-IR. These faint micro-Jansky sources have H > 20.5 and K > 19.5 - 20. The interpretation of these optically faint, micro-Jansky radio sources is that they are potentially candidates of dust enshrouded starburst galaxies beyond a redshift of 1. Our 1.4GHz detection threshold of 40muJy (5sigma) sets the minimum limit of star formation activities of - 200M(circle dot)/yr at z - 1.5, which can be probed by our deep VLA data. An increasing number of recent observations [2,4] support our speculation that optically faint, micro-Jansky radio sources are potential dusty starbursts at high redshift. This may be the most efficient way of detecting a large sample of sub-mm sources with the current available instruments.