[O II] as a star formation rate indicator

We investigate the [O II] emission line as a star formation rate (SFR) indicator using integrated spectra of 97 galaxies from the Nearby Field Galaxies Survey (NFGS). The sample includes all Hubble types and contains SFRs ranging from 0.01 to 100 M-. yr(-1). We compare the Kennicutt [O II] and Halpha SFR calibrations and show that there are two significant effects that produce disagreement between SFR([ O II]) and SFR(Halpha): reddening and metallicity. Differences in the ionization state of the interstellar medium do not contribute significantly to the observed difference between SFR([ O II]) and SFR(Halpha) for the NFGS galaxies with metallicities log (O/H) + 12 greater than or similar to 8.5. The Kennicutt [ O II]-SFR relation assumes a typical reddening for nearby galaxies; in practice, the reddening differs significantly from sample to sample. We derive a new SFR([ O II]) calibration that does not contain a reddening assumption. Our new SFR([ O II]) calibration also provides an optional correction for metallicity. Our SFRs derived from [ O II] agree with those derived from Halpha to within 0.03 - 0.05 dex. We show that the reddening, E( B - V), increases with intrinsic (i.e., reddening-corrected) [ O II] luminosity for the NFGS sample. We apply our SFR([ O II]) calibration with metallicity correction to two samples: high-redshift 0.8 < z < 1.6 galaxies from the NICMOS Halpha survey and 0.5 < z < 1.1 galaxies from the Canada-France Redshift Survey. The SFR([ O II]) and SFR(Halpha) for these samples agree to within the scatter observed for the NFGS sample, indicating that our SFR([ O II]) relation can be applied to both local and high-z galaxies. Finally, we apply our SFR([ O II]) to estimates of the cosmic star formation history. After reddening and metallicity corrections, the star formation rate densities derived from [ O II] and Halpha agree to within similar to 30%.