arXiv:astro-ph/0109143v1 10 Sep 2001 THE NATURE OF THE FAINT RADIO POPULATION I.PR A N D O N I 1; 2 ,L.G R E G O R IN I 1; 2 ,P.PA R M A 2 ,H .R .de RU IT ER 3; 2 ,G.VETTOLANI 2 , M .W IERINGA 4 ,R.EKERS 4 1 U niversita degliStudidiB ologna, B ologna,Italy 2 Istituto diRadioastronom ia delC N R ,B ologna,Italy 3 O sservatorio A stronom ico diB ologna,B ologna,Italy 4 A ustralia Telescope N ationalFacility,Sydney,A ustralia A bstract. W e presentthe results obtained so far from the opticalfollow-up ofthe ATESP sub-m Jy radio survey.T heAT ESP resultsare then com pared w ith the onesobtained from other deep radio sam plesand we show how the existing discrepanciesaboutthe nature ofthe faint radio population can be explained in term sofselection e ects. 1 Introduction R adio source counts derived from deep 1.4 GHz surveysshow a sudden steepening below 1 mJy (e.g. [15],[11]),which is interpreted as the result ofthe em ergence ofa new population ofradio sourcesdi erentin nature from the oneswhich dom inate athigher ux densities(e.g. classicalradio galaxiesand Q SO s). N evertheless,despite thelarge observationale orts,thephysicaland evolutionary properties ofthis population are stillpoorly understood. Today we know thatthe faintradio population is a m ixture ofseveraltypes ofobjects (faintAG Ns, norm alspirals and ellipticals,starburst galaxies),but the relative im portance ofthe di erent classes is stilldebated and very little is know n abouttheir redshiftdistribution and lum inosity properties. T he cause ofitistwofold. First,faintradio sam plesare sm all,w ith sky coverages typically going from a fraction to a few square degrees. T he statistics available istherefore quite poor.Secondly and m ore im portantly,the opticalfollow -up ofthese sam ples is strongly incom plete. Typically 50% 60% of the radio sourcesare identi ed on opticalim ages,while only 20% have spectroscopicinform ation. E xceptions are the Jy sample taken on the HDF North (see [13])wherethe identication rate is 80% and theM arano Field (M F)sub-m Jy sam ple(see[4])with spectroscopy availablefor45% ofthe sources. In both cases,though,the num bers involved are very sm all. T hism eansthatthe resultsdraw n from radio-opticalstudiesofthe faintradio population are based on lim ited opticalfollow -up and biased by the factthatonly the optically brightestsources have spectral information available. It is therefore clear that larger deep radio samples with possibly complete opticalfollow-up are strongly needed in orderto fully assessthe nature and evolution ofthe faint radiopopulation. 2 The ATESP R adio Survey and its O pticalFollow -up W ith thisin m ind,wetook advantageofthem osaicobservingm odecapability oftheAustralia Tele- scope Com pactArray (ATCA)to deeply im age at1.4 GHz the entire region previously covered by the E SO Slice P roject (E SP ) redshift survey (see [14]for a fulldescription). W ehaveproduced 16 radio m osaicswith uniform noiseof 79 Jyand spatialresolution of 8 00 14 00 (see [9]for details). The sky coverage ofthe ATESP survey is26 square degrees,i.e. one order of m agnitude largerthan any otherprevioussub-m Jy sam ple.From the radio im ageswe have extracted a catalogueof2967 radio sources,com pletedown to S 0 : 5 m Jy (see[10]). T he ESP redshiftsurvey ([14])provided usw ith spectroscopy inform ation fora sam ple of3342 galaxies com plete down to b j 19 : 4. Such data allowed usto identify about10% ofthe AT ESP radio sources. T he typicaldepth of the ESP survey (0 <z< 0 : 3)is z 0 : 1. Such sam ple istherefore very well suited to assess the localradio-opticalproperties ofthe faint radio population. In Fig.1 w e show the