On the limitations of H alpha luminosity as a star formation tracer in spatially resolved observations

This study examines the limitations of H$\alpha$ luminosity as a tracer of star formation rates (SFR) in spatially resolved observations. We carry out high-resolution simulations of a Milky Way-like galaxy including both supernova and photoionization feedback, and from these we generate synthetic H$\alpha$ emission maps that we compare to maps of the true distribution of young stellar objects (YSOs) on scales from whole-galaxy to individual molecular clouds ($\lesssim 100$ pc). Our results reveal significant spatial mismatches between H$\alpha$ and true YSO maps on sub-100 pc scales, primarily due to ionizing photon leakage, with a secondary contribution from young stars drifting away from their parent molecular clouds. On small scales these effect contribute significantly to the observed anti-correlation between gas and star formation, such that there is noticeably less anti-correlation if we replace an H$\alpha$-based star formation map with a YSO-based one; this in turn implies that previous studies have underestimated the time it takes for young stars to disperse their parent molecular clouds. However, these effects are limited in dense regions with hydrogen columns $N_\mathrm{H} > 3 \times 10^{21}$ cm$^{-2}$, where the H$\alpha$- and YSO-based SFR maps show better agreement. Based on this finding we propose a calibration model that can precisely measure the SFR of large molecular clouds (mean radius > 100 pc) with a combination of H$\alpha$ and CO observations, which provides a foundation for future study of star formation processes in extragalactic molecular clouds.