We present a sample of 485 photometrically identified Type Ia supernova candidates mined from the first three years of data of the CFHT SuperNova Legacy Survey (SNLS). The images were submitted to a deferred processing independent of the SNLS real-time detection pipeline. Light curves of all transient events were reconstructed in the g_M, r_M, i_M and z_M filters and submitted to automated sequential cuts in order to identify possible supernovae. Pure noise and long-term variable events were rejected by light curve shape criteria. Type Ia supernova identification relied on event characteristics fitted to their light curves assuming the events to be normal SNe Ia. The light curve fitter SALT2 was used for this purpose, assigning host galaxy photometric redshifts to the tested events. The selected sample of 485 candidates is one magnitude deeper than that allowed by the SNLS spectroscopic identification. The contamination by supernovae of other types is estimated to be 4%. Testing Hubble diagram residuals with this enlarged sample allows us to measure the Malmquist bias due to spectroscopic selections directly. The result is fully consistent with the precise Monte Carlo based estimate used to correct SN Ia distance moduli in the SNLS 3-year cosmological analyses. This paper demonstrates the feasibility of a photometric selection of high redshift supernovae with known host galaxy redshifts, opening interesting prospects for cosmological analyses from future large photometric SN Ia surveys.
Read full abstract