Resolved Gas Kinematics in a Sample of Low-Redshift High Star-Formation Rate Galaxies

Abstract

We have used integral field spectroscopy of a sample of six nearby (z ∼ 0.01-0.04) high star-formation rate (SFR ∼ 10-40 M yr −1 ) galaxies to investigate the relationship between local velocity dispersion and star-formation rate on sub-galactic scales. The low-redshift mitigates, to some extent, the effect of beam smearing which artificially inflates the measured dispersion as it combines regions with different line-of-sight velocities into a single spatial pixel. We compare the parametric maps of the velocity dispersion with the Hα flux (a proxy for local star-formation rate), and the velocity gradient (a proxy for the local effect of beam smearing). We find, even for these very nearby galaxies, the Hα velocity dispersion correlates more strongly with velocity gradient than with Hα flux-implying that beam smearing is still having a significant effect on the velocity dispersion measurements. We obtain a first-order non parametric correction for the unweighted and flux weighted mean velocity dispersion by fitting a 2D linear regression model to the spaxel-byspaxel data where the velocity gradient and the Hα flux are the independent variables and the velocity dispersion is the dependent variable; and then extrapolating to zero velocity gradient. The corrected velocity dispersions are a factor of ∼1.3-4.5 and ∼1.3-2.7 lower than the uncorrected flux-weighted and unweighted mean line-of-sight velocity dispersion values, respectively. These corrections are larger than has been previously cited using disc models of the velocity and velocity dispersion field to correct for beam smearing. The corrected flux-weighted velocity dispersion values are σ m ∼ 20-50 km s −1 .

The SAMI Galaxy Survey: Bayesian inference for gas disc kinematics using a hierarchical Gaussian mixture model

“…This expands the previous method to include estimates for the boundary pixels. We also note that within the central pixels the division by 2∆y was omitted previously by Varidel et al (2016). Strictly speaking, this is incorrect as the gradient will be over-estimated by a factor of 2.…”

“…Another approach to correct for the effects of beam smearing on the observed velocity dispersion is to perform corrections based on the local velocity gradient (v grad ). Varidel et al (2016) proposed calculating the local velocity gradient using a finite-difference scheme and then performed a regression analysis to estimate the observed velocity dispersion when the local velocity gradient is zero. Zhou et al (2017) and Federrath et al (2017b) have also used the finite-difference scheme method to remove spaxels where the velocity gradient is much greater than the observed velocity dispersion.…”

“…For the parametric regression estimates we fit a cubic to σ v vs. v grad and then estimated the line at v grad = 0 km s −1 arcsec −1 . We fit a cubic instead of a first-order line in contrast to Varidel et al (2016) as there were clear residuals observed by-eye in the linear and quadratic fits to the data. This method suffered from over-estimates of the mean velocity dispersion similar to that observed using the estimates in regions where σ v v grad FWHM PSF .…”

“…One relevant claim has been that star-formation feedback processes play an important role as a driver of gas turbulence across epochs (Green et al 2010(Green et al , 2014. In contrast, there have been several studies of the localised star-formation rate and gas turbulence in nearby galaxies which have not found a significant correlation (Varidel et al 2016;Zhou et al 2017). Another recent claim has been that gas turbulence may be driven by the interaction between clumps and the interstellar medium (Oliva-Altamirano et al 2018).…”

“…Another approach is to apply corrections to the observed velocity dispersion as a function of properties that exacerbate the effect of beam smearing such as the seeing width and rotational velocity (Johnson et al 2018). The local velocity gradient (Varidel et al 2016) has also been used to ignore spaxels with high local velocity gradient (Zhou et al 2017;Federrath et al 2017b) as well as provide corrections for the global (Varidel et al 2016) and local velocity dispersion (Oliva-Altamirano et al 2018).…”