We use a large data-set of realistic synthetic observations (PaperI) to assess how observational techniques affect the measurement of physical properties of star-forming regions. In this paper (PaperII), we explore the reliability of the measured total gas mass, dust surface density and dust temperature maps derived from modified blackbody fitting of synthetic Herschel observations. We found from our pixel-by-pixel analysis of the measured dust surface density and dust temperature a worrisome error spread especially close to star-formation sites and low-density regions, where for those contaminated pixels the surface densities can be under/overestimated by up to three orders of magnitude. In light of this, we recommend to treat the pixel-based results from this technique with caution in regions with active star formation. In regions of high background typical in the inner Galactic plane, we are not able to recover reliable surface density maps of individual synthetic regions, since low-mass regions are lost in the FIR background. When measuring the total gas mass of regions in moderate background, we find that modified blackbody fitting works well (absolute error:+9%;-13%) up to 10kpc distance (errors increase with distance). Commonly, the initial images are convolved to the largest common beam-size, which smears contaminated pixels over large areas. The resulting information loss makes this commonly-used technique less verifiable as now chi^2-values cannot be used as a quality indicator of a fitted pixel. Our control measurements of the total gas mass (without the step of convolution to the largest common beam size) produce similar results (absolute error:+20%;-7%) while having much lower median errors especially for the high-mass stellar feedback phase. In upcoming papers (III&IV) we test the reliability of measured star-formation rate with direct and indirect techniques.
Download