The Star Formation History of Galaxies Measured from Individual Pixels. I. The Hubble Deep Field North

We analyze the photometric information contained in individual pixels ofgalaxies in the Hubble Deep Field North (HDFN) using a new technique,_pixel-z_, that combines predictions of evolutionary synthesis models withphotometric redshift template fitting. Each spectral energy distributiontemplate is a result of modeling of the detailed physical processes affectinggas properties and star formation efficiency. The criteria chosen to generatethe SED templates is that of sampling a wide range of physical characteristicssuch as age, star formation rate, obscuration and metallicity. A key feature ofour method is the sophisticated use of error analysis to generate error mapsthat define the reliability of the template fitting on pixel scales and allowfor the separation of the interplay among dust, metallicity and star formationhistories. This technique offers a number of advantages over traditionalintegrated color studies. As a first application, we derive the star formationand metallicity histories of galaxies in the HDFN. Our results show that thecomoving density of star formation rate, determined from the UV luminositydensity of sources in the HDFN, increases monotonically with redshift out to atleast redshift of 5. This behavior can plausibly be explained by a smoothincrease of the UV luminosity density with redshift coupled with an increase inthe number of star forming regions as a function of redshift. We also find thatthe information contained in individual pixels in a galaxy can be linked to itsmorphological history. Finally, we derive the metal enrichment rate history ofthe universe and find it in good agreement with predictions based on theevolving HI content of Lyman-alpha QSO absorption line systems.