The CANADA-FRANCE REDSHIFT SURVEY XIII: The luminosity density and star-formation history of the Universe to z ~ 1

The comoving luminosity density of the Universe is estimated from the CFRS faint galaxy sample in three wavebands (2800A, 4400A and 1 micron) over the redshift range 0 < z < 1. In all three wavebands, the comoving luminosity density increases markedly with redshift. For a (q_0 = 0.5, Omega = 1.0) cosmological model, the comoving luminosity density increases as $(1+z)^{2.1 \pm 0.5}$ at 1 micron, as $(1+z)^{2.7 \pm 0.5}$ at 4400A and as $(1+z)^{3.9 \pm 0.75}$ at 2800A, these exponents being reduced by 0.43 and 1.12 for (0.05,0.1) and (-0.85,0.1) cosmological models respectively. The variation of the luminosity density with epoch can be reasonably well modelled by an actively evolving stellar population with a Salpeter initial mass function (IMF) extending to 125 M_sun, a star-formation rate declining with a power 2.5, and a turn-on of star-formation at early epochs. A Scalo (1986) IMF extending to the same mass limit produces too many long-lived low mass stars. This rapid evolution of the star-formation rate and comoving luminosity density of the Universe is in good agreement with the conclusions of Pei and Fall (1995) from their analysis of the evolving metallicity of the Universe. One consequence of this evolution is that the physical luminosity density at short wavelengths has probably declined by two orders of magnitude since z ~ 1.