http://arxiv.org/abs/2305.07874
The growing range of automated algorithms for the identification of molecular clouds and clumps in large observational datasets has prompted the need for the direct comparison of these procedures. However, these methods are complex and testing for biases is often problematic: only a few of them have been applied to the same data set or calibrated against a common standard. We compare the Fellwalker method, a widely used watershed algorithm, to the more recent Spectral Clustering for Interstellar Molecular Emission Segmentation (SCIMES). SCIMES overcomes sensitivity and resolution biases that plague many friends-of-friends algorithms by recasting cloud segmentation as a clustering problem. Considering the \ce{^{13}CO}/\ce{C^{18}O} ($J = 3 – 2$) Heterodyne Inner Milky Way Plane Survey (CHIMPS) and the CO High-Resolution Survey (COHRS), we investigate how these two different approaches influence the final cloud decomposition. Although the two methods produce largely similar statistical results over the CHIMPS dataset, FW appears prone to over-segmentation, especially in crowded fields where gas envelopes around dense cores are identified as adjacent, distinct objects. FW catalogue also includes a number of fragmented clouds that appear as different objects in a line-of-sight projection. In addition, cross-correlating the physical properties of individual sources between catalogues is complicated by different definitions, numerical implementations, and design choices within each method, which make it very difficult to establish a one-to-one correspondence between the sources.
R. Rani, T. Moore, D. Eden, et. al.
Tue, 16 May 23
64/83
Comments: accepted MNRAS