The Impact of Line Misidentification on Cosmological Constraints from Euclid and other Spectroscopic Galaxy Surveys

We perform forecasts for how baryon acoustic oscillation (BAO) scale and redshift-space distortion (RSD) measurements from future spectroscopic emission line galaxy (ELG) surveys such as Euclid are degraded in the presence of spectral line misidentification. Using analytic calculations verified with mock galaxy catalogs from log-normal simulations we find that constraints are degraded in two ways, even when the interloper power spectrum is modeled correctly in the likelihood. Firstly, there is a loss of signal-to-noise ratio for the power spectrum of the target galaxies, which propagates to all cosmological constraints and increases with contamination fraction, $f_c$. Secondly, degeneracies can open up between $f_c$ and cosmological parameters. In our calculations this typically increases BAO scale uncertainties at the 10-20% level when marginalizing over parameters determining the broadband power spectrum shape. External constraints on $f_c$, or parameters determining the shape of the power spectrum, for example from cosmic microwave background (CMB) measurements, can remove this effect. There is a near-perfect degeneracy between $f_c$ and the power spectrum amplitude for low $f_c$ values, where $f_c$ is not well determined from the contaminated sample alone. This has the potential to strongly degrade RSD constraints. The degeneracy can be broken with an external constraint on $f_c$, for example from cross-correlation with a separate galaxy sample containing the misidentified line, or deeper sub-surveys.