2005A&A...429..851J

We study the two main corrections generally applied to narrow-band Hα fluxes from galaxies in order to convert them to star formation rates, namely for [NII] contamination and for extinction internal to the galaxy. From an imaging study using carefully chosen narrow-band filters, we find the [NII] and Hα emission to be differently distributed. Nuclear measurements are likely to overestimate the contribution of [NII] to total narrow-band fluxes. We find that in most star formation regions in galaxy disks the [NII] fraction is small or negligible, whereas some galaxies display a diffuse central component which can be dominated by [NII] emission. We compare these results with related studies in the literature, and consider astrophysical explanations for variations in the [NII]/Hα ratio, including metallicity variations and different excitation mechanisms. We proceed to estimate the extinction towards star formation regions in spiral galaxies, firstly using Brγ/Hα line ratios. We find that extinction values are larger in galaxy nuclei than in disks, that disk extinction values are similar to those derived from optical emission-line studies in the literature, and that there is no evidence for heavily dust-embedded regions emerging in the near-IR, which would be invisible at Hα. The numbers of galaxies and individual regions detected in Brγ are small, however, and we thus exploit optical emission line data from the literature to derive global Hα extinction values as a function of galaxy type and inclination. In this part of our study we find only a moderate dependence on inclination, consistent with broad-band photometric studies, and a large scatter from galaxy to galaxy. Typical extinctions are smaller for late-type dwarfs than for spiral types. Finally, we show that the application of the type-dependent extinction corrections derived here significantly improves the agreement between star formation rates calculated using Hα fluxes and those from far-infrared fluxes as measured by IRAS. This again supports the idea that heavily dust-embedded star formation, which would be underestimated using the Hα technique, is not a dominant contributor to the total star formation rate of most galaxies in the local Universe.