Uncertainties in the Deprojection of the Observed Bar Properties
In observations, it is important to deproject the two fundamental quantities characterizing a bar, i.e., its length ($a$) and ellipticity ($e$), to face-on values before any careful analyses. However, systematic estimation on the uncertainties of the commonly used deprojection methods is still lacking. Simulated galaxies are well suited in this study. We project two simulated barred galaxies onto a 2D plane with different bar orientations and disk inclination angles ($i$). Bar properties are measured and deprojected with the popular deprojection methods in the literature. Generally speaking, deprojection uncertainties increase with increasing $i$. All the deprojection methods behave badly when $i$ is larger than $60^\circ$, due to vertical thickness of the bar. Thus, future statistical studies of barred galaxies should exclude galaxies more inclined than $60^\circ$. At moderate inclination angles ($i\leq60^\circ$), 2D deprojection methods (analytical and image stretching) and Fourier-based methods (Fourier decomposition and bar-interbar contrast) perform reasonably well with uncertainties $\sim10\%$ in both the bar length and ellipticity. Whereas the uncertainties of the 1D analytical deprojection can be as high as $100\%$ in certain extreme case. We find that different bar measurement methods show systematic differences in the deprojection uncertainties. We further discuss the deprojection uncertainty factors with the emphasis on the most important one, i.e., the 3D structure of the bar itself. We construct two triaxial toy bar models that can qualitatively reproduce the results of the 1D and 2D analytical deprojections; they confirm that the vertical thickness of the bar is the main source of uncertainties.