COMPARING ULTRAVIOLET-AND INFRARED-SELECTED STARBURST GALAXIES IN DUST OBSCURATION AND LUMINOSITY

We present samples of starburst galaxies that represent the extremes discovered with infrared and ultraviolet observations, including 25 Markarian galaxies, 23 ultraviolet-luminous galaxies discovered with GALEX, and the 50 starburst galaxies having the largest infrared/ultraviolet ratios. These sources have z < 0.5 and cover a luminosity range of similar to 10(4). Comparisons between infrared luminosities determined with the 7.7 mu m polycyclic aromatic hydrocarbon feature and ultraviolet luminosities from the stellar continuum at 153 nm are used to determine obscuration in starbursts and dependence of this obscuration on infrared or ultraviolet luminosity. A strong selection effect arises for the ultraviolet- selected samples: the brightest sources appear bright because they have the least obscuration. Obscuration correction for the ultraviolet- selected Markarian+GALEX sample has the form log[UV(intrinsic)/UV(observed)] = 0.07(+/- 0.04)M(UV) + 2.09 +/- 0.69 but for the full infrared-selected Spitzer sample is log[UV(intrinsic)/UV(observed)] = 0.17(+/- 0.02)M(UV) + 4.55 +/- 0.4. The relation of total bolometric luminosity L-ir to M(UV) is also determined for infrared-selected and ultraviolet-selected samples. For ultraviolet-selected galaxies, log (Lir) = -(0.33 +/- 0.04) M(UV) + 4.52 +/- 0.69. For the full infrared-selected sample, log L-ir = -(0.23 +/- 0.02) M(UV) + 6.99 +/- 0.41, all for L-ir in L-circle dot and M(UV) the AB magnitude at rest frame 153 nm. These results imply that obscuration corrections by factors of 2-3 determined from reddening of the ultraviolet continuum for Lyman break galaxies with z > 2 are insufficient, and should be at least a factor of 10 for M(UV) similar to -17, with decreasing correction for more luminous sources.