Gaussian Decomposition of HI Surveys - II. Separation of Problematic Gaussians

Abstract

We have analyzed the Gaussian decomposition of the Leiden/Dwingeloo Survey (LDS) of galactic neutral hydrogen for the presence of Gaussians probably not directly related to galactic HI emission. It is demonstrated that at least three classes of such components can be distinguished. The narrowest Gaussians mostly represent stronger random noise peaks in profiles and some still uncorrected radio-interferences. Many of slightly wider weak Gaussians are caused by increased uncertainties near the profile edges and with the still increasing width the baseline problems become dominating among weak components. Statistical criteria are given for separation of the parameter space regions, most likely populated with the problematic components from those where the Gaussians are with higher probability describing the actual Milky Way HI emission. The same analysis is applied to the Leiden/Argentina/Bonn survey (LAB). It is demonstrated that the selection criteria for dividing the parameter space are to a great extent independent of the particular survey in use. The presence of the baseline problems in the LDS is indicated by the peculiarities of the distribution of the widest Gaussians in the sky. A similar plot for the northern part of the LAB demonstrates considerably lower numbers of spurious components, but there are still problems with the southern part of the LAB. The strange characteristics of the observational noise in the southern part of the LAB are pointed out.