2011MNRAS.415.3616D

We analyse the reliability of oxygen abundances and ionization parameters obtained from a number of diagnostic diagrams. To do this, we used the literature to compile the observational emission-line intensities and oxygen abundances of 446 star-forming regions whose O/H abundances were determined by direct estimation of the electron temperature. These compiled abundances were compared with the values calculated in this work using various diagnostic diagrams in combination with results from a grid of photoionization models. We found that the [O III]/[O II] versus [N II]/[O II], [O III]/Hβ versus [N II]/[O II] and ([O III]/Hβ)/([N II]/Hα) versus [S II]/[S III] diagnostic diagrams gave O/H values close to those obtained using the electron temperature, with differences of about 0.04 dex and a dispersion of about 0.3 dex. Similar results were obtained by detailed models, but with a dispersion of 0.08 dex. The origin of the dispersion found with the use of diagnostic diagrams is probably the differences between the real N/O–O/H relation of the sample and the one assumed in the models. This is confirmed by the use of detailed models that do not have a fixed N/O–O/H relation. We found no correlation between the ionization parameter and metallicity for the objects of our sample. We conclude that the combination of two line ratios predicted by photoionization models, one sensitive to the metallicity and the other sensitive to the ionization parameter, which takes into account the physical conditions of star-forming regions, gives O/H estimates close to the values derived using direct detections of electron temperature. Galaxy Zoo: bar lengths in local disc galaxies