Oxygen abundance determination in HII regions: The strong line intensities-abundance calibration revisited

The problem of oxygen abundance determination in H II regions based on the ''strong oxygen line intensities oxygen abundance'' empirical calibration is revisited. A compilation of spectroscopic data of H II regions in spiral and irregular galaxies with a measured [O III] lambda 4363 line intensity has been carried out, resulting in a sample containing more than 700 individual measurements. Methods are devised to select out only high-precision measurements from that original sample. T-e-based oxygen abundances have been recomputed in the same way for all H II regions with high-precision measurements. That sample of T-e abundances is used to recalibrate the empirical relations between the oxygen abundance and the strong oxygen line intensities for both high (the upper branch of the O/H-R-23 diagram) and low (the lower branch) metallicities, within the framework of the P method, where P is the excitation parameter. Concerning high-metallicity H II regions, an alternative way for deriving oxygen abundances using only measurements of the strong nebular oxygen lines is proposed. The method is based on a tight correlation between the flux in the auroral [O III] lambda 4363 line and the fluxes in the nebular [O II] lambda lambda 3727, 3729 and [O III] lambda lambda 4959, 5007 lines, called the ff relation. This relation is also used to select out high-metallicity H II regions with high-precision (O/H)T-e measurements. In contrast to previous work, the new upper branch P calibration is based only on (O/H)T-e abundances. It is found that (O/H)p abundances usually agree well with the (O/H)(ff) abundances, although faint (log R23 less than or similar to - 0.5) low-excitation (P less than or similar to 0.25) H II regions may show systematic differences that can be as large as similar to 0.1 dex. As for the newly derived low-metallicity P calibration, it is shown to be robust. The calibrations derived from the sample containing all (O/H)T-e abundance determinations and from that containing only recent (since 1995) measurements are found to be in very good agreement. For both low- and high-metallicity H II regions, the new calibration gives (O/H)(p) abundances that agree with (O/ H)(p) abundances to within 0.1 dex.