Are supernovae driving turbulence in the solar neighborhood?

Abstract

Turbulence plays an important role in shaping the interstellar medium, and strongly influences star formation. We aim to identify the physical processes capable of sustaining HI turbulence in the solar neighborhood. We compare recent HI line-of-sight velocity observations within a volume of radius 70-500 pc centered on the Sun with a suite of 1 kpc numerical simulations that include two distinct turbulent drivers: (i) supernova (SN) feedback and (ii) imposed large-scale turbulent forcing. For each simulation, we construct synthetic sky maps that closely mimic the observational one, allowing for a consistent comparison between the simulations and the observational data. HI observations show a median velocity dispersion of 11.1 km s-1 in the solar neighborhood. SN-driven simulations systematically underpredict this value, yielding dispersions in the range 4.9-6.7 km s-1. Simulations with strong enough large-scale forcing can reproduce not only the median observed velocity dispersion, but also the observed velocity distribution.

Figures (8)

• Figure 1: Probability density functions of mass-weighted line-of-sight velocity dispersions for all simulations. Each curve shows the distribution of dispersion values obtained from the different masked sky maps for a given simulation. Dashed curves correspond to the large-scale–forced TURB runs, while dotted curves denote the supernova-driven simulations (SN series and TIGRESS-MHD4pc). The dot-dashed vertical line marks the median H i velocity dispersion, $\sigma_{50} = 11.1^{+0.9}_{-0.3}~\mathrm{km\,s^{-1}}$, derived from the Monte Carlo sampling of the line-of-sight velocities reported by Soler_2025, using the Reid_2019 Galactic rotation model parameters in a volume spanning 70--500 pc around the Sun. The gray shaded vertical band represents the corresponding 25th--75th percentile range. Velocity dispersions are computed after applying a velocity threshold of $|v_{\mathrm{los}}| \le 30~\mathrm{km\,s^{-1}}$.
• Figure 2: Observed and simulated line-of-sight velocity maps before applying the observational mask.
• Figure 3: Masked synthetic sky maps obtained by applying the observational sampling pattern at different rotation angles to the full synthetic map shown in Fig. \ref{['fig:cake_full']} (b).
• Figure 4: Evolution of the velocity PDF as successive numerical processing steps are applied to one TURB-15 map. The gray histogram shows the H i PDF from Soler_2025 ($70$--$500$ pc, $|v_{\rm los}| \le 40~\mathrm{km\,s^{-1}}$). Blue histograms correspond to (A) the intrinsic full-resolution 3D simulation, (B) the 2D $(x,y)$ map obtained by averaging along $z$, (C) the synthetic sky projection (4 pc radial, $10^\circ$ angular sampling), and (D) the same projection after applying one observational mask.
• Figure 5: Evolution of the mass-weighted line-of-sight velocity dispersion as a function of the imposed LOS–velocity threshold for both simulations and observations. Colored curves show the simulated models, averaged over the corresponding sets of synthetic sky maps, with supernova-driven runs shown as colored dotted lines and large-scale turbulent-forcing runs as colored dashed lines. Black curves correspond to the observational H i data from Soler_2025: solid for the full dataset, dotted for the outer Galaxy, and dash-dotted for the inner Galaxy.