Spectral classification of emission-line galaxies from the Sloan Digital Sky Survey I. An improved classification for high-redshift galaxies

Aims. We study the spectral classification of emission-line galaxies as starforming galaxies or active galactic nuclei (AGNs). With the high-quality data from the Sloan Digital Sky Survey (SDSS) we define an improved classification to be used for high-redshift galaxies. Methods. We classify emission-line galaxies of the SDSS according to the latest standard recipe using [OIII]lambda 5007, [NII]lambda 6584, [SII]lambda 6717+6731, H alpha, and H beta emission lines. We obtain four classes: starforming galaxies, Seyfert 2, LINERs, and composites. We then examine where these galaxies fall in the blue diagram used at high redshift (i.e. log([OIII].5007/H beta) vs. log([OII]lambda lambda 3726+3729/H beta). Results. We define new improved boundaries in the blue diagram for starforming galaxies, Seyfert 2, LINERs, SF/Sy2, and SF-LIN/comp classes. We maximize the success rate to 99.7% for the detection of starforming galaxies to 86% for the Seyfert 2 (including the SF/Sy2 region) and to 91% for the LINERs. We also minimize the contamination to 16% in the region of starforming galaxies. We cannot reliably separate composites from starforming galaxies and LINERs, but we define an SF-LIN/comp region where most of them fall (64%).