2013ApJ...773...88S

Neutral hydrogen (H I) velocity dispersions are believed to be set by turbulence in the interstellar medium (ISM). Although turbulence is widely believed to be driven by star formation, recent studies have shown that this driving mechanism may not be dominant in regions of low star formation surface density (Σ_SFR), such those as found in dwarf galaxies or the outer regions of spirals. We have generated average H I line profiles in a number of nearby dwarfs and low-mass spirals by co-adding H I spectra in subregions with either a common radius or Σ_SFR. We find that the individual spatially resolved "superprofiles" are composed of a central narrow peak (∼5-15 km/s) with higher velocity wings to either side, similar to their global counterparts as calculated for the galaxy as a whole. Under the assumption that the central peak reflects the H I turbulent velocity dispersion, we compare measures of H I kinematics determined from the superprofiles to local ISM properties, including surface mass densities and measures of star formation. The shape of the wings of the superprofiles do not show any correlation with local ISM properties, which indicates that they may be an intrinsic feature of H I line-of-sight spectra. On the other hand, the H I velocity dispersion is correlated most strongly with baryonic and H I surface mass density, which points toward a gravitational origin for turbulence, but it is unclear which, if any, gravitational instabilities are able to operate efficiently in these systems. Star formation energy is typically produced at a level sufficient to drive H I turbulent motions at realistic coupling efficiencies in regimes where Σ_SFR ≳ 10^–4 M_☉/yr/kpc2, as is typically found in inner spiral disks. At low star formation intensities, on the other hand, star formation cannot supply enough energy to drive the observed turbulence, nor does it uniquely determine the turbulent velocity dispersion. Nevertheless, even at low intensity, star formation does appear to provide a lower threshold for H I velocity dispersions. We find a pronounced decrease in coupling efficiency with increasing Σ_SFR, which would be consistent with a picture where star formation couples to the ISM with constant efficiency, but that less of that energy is found in the neutral phase at higher Σ_SFR. We have examined a number of potential drivers of H I turbulence, including star formation, gravitational instabilities, the magneto-rotational instability, and accretion-driven turbulence, and found that, individually, none of these drivers is capable of driving the observed levels of turbulence in the low Σ_SFRregime. We discuss possible solutions to this conundrum.