3D CMZ V: A new orbital model of our Galaxy's Center, informed by data across the electromagnetic spectrum

Abstract

The 3D structure of The Milky Way's Central Molecular Zone (CMZ) informs our understanding of star formation cycles, black hole accretion, and the evolution of galactic nuclei. However, a comprehensive 3D model has remained elusive, as no singular dataset nor theory contains the requisite information to describe the orbital motion of the gas. We implement a Bayesian framework to flexibly combine datasets across the electromagnetic spectrum for molecular clouds in our CMZ catalog. We develop near/far metrics for each dataset, including dust extinction, absorption, stellar densities, X-ray echoes, and proper motions; and report a posterior positional probability density function (PPDF) for each cloud. We then use the posterior PPDF distributions for all CMZ clouds to search for a best fitting x$_2$ orbit. We find that no single orbit is a perfect fit, but the structure can overall be represented by nested x$_2$ orbits, with major axes ranging from about $72 < a < 146$ pc. We also present projected line of sight distance estimates for all 31 clouds in the catalog. Our results highlight asymmetries along the line of sight, with most clouds lying on the near side of the Galactic Center, and agree overall with current near/far assumptions for most CMZ clouds, including those in the Sgr A region, which may be much closer to the center. We conclude that the CMZ can be well-described by x$_2$ orbital families, and that the overall gas distribution is more complex than a single closed or open elliptical orbit.

Figures (12)

• Figure 1: We present a method to synthesize multi wavelength datasets to obtain most likely near/far positions of clouds in the CMZ, and find a best-fitting elliptical orbit. Our methodology follows the above schematic from left to right: 1) we collect both near/far and line of sight distance constraints from a variety of methods across the electromagnetic spectrum (Section \ref{['sec:data_and_methods_EM']}), 2) we build a Bayesian framework to combine these different data on a normalized near-to-far scale, producing positional probability density functions which peak at the most likely position of a given cloud (Section \ref{['subsec:bayseian_posteriors']}), 3) Find a best fitting closed elliptical x$_2$ orbit using each cloud's PPV location and NF distribution (Section \ref{['subsec:fitting_procedures']}), 4) Use the best fit orbital model to present a preliminary top-down projection of the CMZ with line-of-sight distance estimates for all CMZ clouds (Section \ref{['sec:disc_relative_positions']}). The black star in panels 3 and 4 indicates the location of SgrA$^{\star}$. The blue and red colors of clouds in panels 3 and 4 indicate likely near or far positions, respectively.
• Figure 1: Most normalized NF distributions can be approximated by a single Gaussian. We highlight four examples of clouds showing the distribution of normalized counts for the flux difference (first column), flux ratio (second column), star count ratio (third column), and absorption methods (fourth column). Each plot shows a histogram of the cloud maps with NF values normalized using Eqn \ref{['eqn:z-scale']} (gray), the median of the normalized map (blue dash-dotted line), and the normalized PPDF prior ultimately used for each method (orange line).
• Figure 1: Prior and posterior PPDFs for all cloud masks in the catalog. Each panel shows priors for different methods as colored lines, and the posterior is indicated by a black solid line. We report the location of the peak ($\mu$, vertical gray dash-dotted line), the relative probability peak (A), and the 68% confidence interval ($\mathrm{CI}_{68}$, horizontal solid gray line), which are used to obtain a posterior PPDF distribution used for the orbital fitting procedure in Section \ref{['subsec:fitting_procedures']}.
• Figure 3: Stellar density maps of the CMZ show dark extinction features that can be used to calculate a star count ratio to determine NF positions of molecular clouds in the catalog. From top to bottom, the panels show: (i) Herschel column density map with black contours and ID numbers correspond to cloud masks summarized in Table \ref{['tab:los_synth_table']}, (ii) stellar density map from Nishiyama2013 and (iii) modeled star count map from Gallego-Cano2020, both of which we convolve and regrid to match the 36$^{\prime\prime}$ Herschel column density map. White areas are masked pixels where the star count is greater than the maximum of the model. Panel (iv) is a cutout of star counts for the molecular clouds; panel (v) is a cutout of modeled star counts; and panel (vi) is a ratio of the star counts to the model map. The blue and red color scales denote more likely near or far side positions, respectively. Clouds containing $> 20\%$ of masked pixels are excluded from analysis. The black star indicates the location of SgrA$^{\star}$.
• Figure 5: The methods used to infer NF positions of individual clouds in the CMZ can be combined in a Bayesian framework to create a posterior distribution of NF likely positions for a given cloud on a normalized scale of -1 (far) to +1 (near). The flux difference (orange dashed line) is used as a likelihood function, and is multiplied by distributions for the flux ratio (green), correlation coefficient (blue), absorption fraction (red), and star count ratios (yellow). A handful of clouds have prior data from X-rays (pink), or stellar kinematics (brown). After taking the product of the priors, we report the location of the peak ($\mu$), the relative probability peak (A), and the 68% confidence interval ($\mathrm{CI}_{68}$), which are used to obtain a Positional PDF posterior distribution (PPDF; black solid line). The peak of the posterior indicates the most likely NF position of the cloud. All distributions are normalized to unit integral probability.