Understanding and reducing statistical uncertainties in nebular abundance determinations

Abstract

Whenever observations are compared to theories, an estimate of the uncertainties associated with the observations is vital if the comparison is to be meaningful. However, many determinations of temperatures, densities and abundances in photoionized nebulae do not quote the associated uncertainty. Those that do typically propagate the uncertainties using analytical techniques which rely on assumptions that generally do not hold. Motivated by this issue, we have developed NEAT (Nebular Empirical Analysis Tool), a new code for calculating chemical abundances in photoionized nebulae. The code carries out an analysis of lists of emission lines using long-established techniques to estimate the amount of interstellar extinction, calculate representative temperatures and densities, compute ionic abundances from both collisionally excited lines and recombination lines, and finally to estimate total elemental abundances using an ionization correction scheme. NEAT uses a Monte Carlo technique to robustly propagate uncertainties from line flux measurements through to the derived abundances. We show that for typical observational data, this approach is superior to analytic estimates of uncertainties. NEAT also accounts for the effect of upward biasing on measurements of lines with low signal to noise, allowing us to accurately quantify the effect of this bias on abundance determinations. We find not only that the effect can result in significant over-estimates of heavy element abundances derived from weak lines, but that taking it into account reduces the uncertainty of these abundance determinations. Finally, we investigate the effect of possible uncertainties in R, the ratio of selective to total extinction, on abundance determinations. We find that the uncertainty due to this parameter is negligible compared to the statistical uncertainties due to typical line flux measurement uncertainties.