EPoS Contribution

Identifying Molecular Cloud Structures in Projected Data: What is Real, and What Isn't?

Christopher Beaumont
University of Hawaii / Harvard, Cambridge, MA, UnitedStates
Molecular clouds are always viewed in projection -- either as two-dimensional maps or as spectral line cubes in position-position-velocity (p-p-v) space. "Closed contour" features in these data do not always correspond to intrinsic structures in the cloud due to the missing line-of-sight information. While well-known, there is no general framework for dealing with this projection problem. We have developed a framework for quantifying the projection problem using hydrodynamic simulations of molecular clouds. Using standard source extraction techniques, we catalog structures both in the 3D density field and in synthetic observations of the simulation. We then study how individual “true” sub-structures project onto the space of the observation and determine their similarity to objects in the catalog of observed structures. This strategy effectively cross-matches the two catalogs, revealing which intrinsic structures have clearly-observed counterparts. By comparing the quality of recovery to properties of the observed structures (e.g. size, flux, shape, and correlation with dense gas tracers), we can infer which structures in observations of real clouds are most likely to be genuine. In many cases, the degree to which we are "fooled" into believing p-p-v structures are real p-p-p objects is disturbing, with as many as 50% of the features in a 13CO map being spurious. The prevalence of artificial structures calls into question analyses of their properties, such as their "boundedness" according to the virial theorem as it is typically applied to spectral-line data.
Alyssa Goodman, Harvard, United States
Stella Offner, Harvard, United States