Improved Measurements of Galaxy Star Formation Stochasticity from the Intrinsic Scatter of Burst Indicators

TL;DR

This paper develops a new method to measure galaxy star formation burstiness by analyzing the intrinsic scatter in burst indicators, accounting for confounding factors like dust extinction variations, and compares models with observations.

Contribution

It introduces an analytical approach to derive intrinsic burstiness from observed scatter, improving upon previous methods by considering $Q_{sg}$ variations.

Findings

Mock and observed burstiness distributions show reasonable agreement.

Systematic differences exist between models and observations in median and scatter.

Intrinsic burstiness is estimated to be between 0.06 and 0.16 dex.

Abstract

Measurements of short-timescale star formation variations (i.e., "burstiness") are integral to our understanding of star formation feedback mechanisms and the assembly of stellar populations in galaxies. We expand upon the work of Broussard et al. (2019) by introducing a new analysis of galaxy star formation burstiness that accounts for variations in the $Q_{sg}=\mathrm{E(B-V)_{stars}}/\mathrm{E(B-V)_{gas}}$ distribution, a major confounding factor. We use Balmer decrements from the MOSFIRE Deep Evolution Field (MOSDEF) survey to measure $Q_{sg}$, which we use to construct mock catalogs from the Santa Cruz Semi-Analytic Models and Mufasa cosmological hydrodynamical simulation based on 3D-HST, Fiber Multi-Object Spectrograph (FMOS)-COSMOS, and MOSDEF galaxies with H$\alpha$ detections. The results of the mock catalogs are compared against observations using the burst indicator $\eta = \log_{10}(\mathrm{SFR_{H\alpha}/SFR_{NUV}})$, with the standard deviation of the $\eta$ distribution indicating burstiness. We find decent agreement between mock and observed $\eta$ distribution shapes; however, the FMOS-COSMOS and MOSDEF mocks show a systematically low median and scatter in $\eta$ in comparison to the observations. This work also presents the novel approach of analytically deriving the relationship between the intrinsic scatter in $\eta$, scatter added by measurement uncertainties, and observed scatter, resulting in an intrinsic burstiness measurement of $0.06-0.16$ dex.