SIMBAD references

We study the reliability of the mass estimates obtained for molecular cloud cores using sub-millimetre and infrared dust emission. We use magnetohydrodynamic simulations and radiative transfer to produce synthetic observations with spatial resolution and noise levels typical of Herschel surveys. We estimate dust colour temperatures using different pairs of intensities, calculate column densities with opacity at one wavelength, and compare the estimated masses with the true values. We compare these results to the case when all five Herschel wavelengths are available. We investigate the effects of spatial variations of dust properties and the influence of embedded heating sources. Wrong assumptions of dust opacity and its spectral index β can cause significant systematic errors in mass estimates. These are mainly multiplicative and leave the slope of the mass spectrum intact, unless cores with very high optical depth are included. Temperature variations bias the colour temperature estimates and, in quiescent cores with optical depths higher than for normal stable cores, masses can be underestimated by up to one order of magnitude. When heated by internal radiation sources, the dust in the core centre becomes visible and the observations recover the true mass spectra. The shape, although not the position, of the mass spectrum is reliable against observational errors and biases introduced in the analysis. This changes only if the cores have optical depths much higher than expected for basic hydrostatic equilibrium conditions. Observations underestimate the value of β whenever there are temperature variations along the line of sight. A bias can also be observed when the true β varies with wavelength. Internal heating sources produce an inverse correlation between colour temperature and β that may be difficult to separate from any intrinsic β(T) relation of the dust grains. This suggests caution when interpreting the observed mass spectra and the spectral indices.