Gaussian Decomposition of λ21-cm Interstellar HI profiles
Following an established protocol of science, that results must be reproducible, we examine the Gaussian fits to Galactic 21-cm emission profiles obtained by two seemingly complementary methods: the semi-automated approach based on the method used by Verschuur (2004) and the automated technique of Nidever et al. (2008). Both methods use data from the Leiden/Argentine/Bonn all-sky survey. The appeal of an automated routine is great, if for no other reason than the time saved over semi-automated fits. The pitfalls, however, are often unanticipated, and the most important aspect of any algorithm is the reproducibility of the results. The comparisons led to the identification of four problems with the Nidever et al. (2008) analysis: (1) different methods of calculating the reduced chi-squared measuring the goodness of fit; (2) an ultra-broad component found bridging the gap between low and intermediate velocity gas; (3) the lack of an imposed spatial coherence allowing different components to appear and disappear in profiles separated by a fraction of a beam width; and (4) multiple, fundamentally different solutions for the profiles at both the North and South Galactic Poles. A two-step method would improve the algorithm, where an automated fit is followed by a quality-assurance, visual inspection. Confirming evidence emerges from this study of a pervasive component with a line width of order 34 km/s, which may be explained by the Critical Ionization Velocity (CIV) of helium. Since the Nidever et al. (2008) paper contains the only result in the refereed literature that contradicts the CIV model, it is important to understand the flaws in the analysis that let to this contradiction.