Synthetic line and continuum observations of simulated turbulent clouds: the apparent widths of filaments.

Filamentary structures are ubiquitous in observations of real molecular clouds, and also in simulations of turbulent, self-gravitating gas. However, making comparisons between observations and simulations is complicated by the difficulty of estimating volume-densities observationally. Here, we have post-processed hydrodynamical simulations of a turbulent isothermal molecular cloud, using a full time-dependent chemical network. We have then run radiative transfer models to obtain synthetic line and continuum intensities that can be compared directly with those observed. We find that filaments have a characteristic width of $\,\sim\!0.1 \, {\rm pc}$, both on maps of their true surface density, and on maps of their $850 \, {\rm \mu m}$ dust-continuum emission, in agreement with previous work. On maps of line emission from CO isotopologues, the apparent widths of filaments are typically several times larger because the line intensities are poorly correlated with the surface density. On maps of line emission from dense-gas tracers such as N$_2$H$^+$ and HCN, the apparent widths of filaments are $\lesssim 0.1 \, {\rm pc}$. Thus, current observations of molecular-line emission are compatible with the universal $0.1 \, {\rm pc}$ filament width inferred from ${\it Herschel}$ observations, provided proper account is taken of abundance, optical-depth, and excitation considerations. We find evidence for $\sim 0.4 \, {\rm km \, s^{-1}}$ radial velocity differences across filaments. These radial velocity differences might be a useful indicator of the mechanism by which a filament has formed or is forming, for example the turbulent cloud scenario modelled here, as against other mechanisms such as cloud-cloud collisions.