On the Significance of Absorption Features in HST/COS Data

TL;DR

This paper develops empirical relations to evaluate the significance of absorption features in HST/COS far-UV spectra, accounting for instrument effects and noise properties, and compares coaddition algorithms' impact on detection limits.

Contribution

It introduces new empirical scaling relations that incorporate COS line spread function wings and non-Poissonian noise, improving absorption feature significance estimates.

Findings

Characterized COS noise properties for the first time.

Coadded data have slightly larger limiting equivalent widths than individual exposures.

Limiting equivalent widths depend on Doppler parameter and coaddition method.

Abstract

We present empirical scaling relations for the significance of absorption features detected in medium resolution, far-UV spectra obtained with the Cosmic Origins Spectrograph (COS). These relations properly account for both the extended wings of the COS line spread function and the non-Poissonian noise properties of the data, which we characterize for the first time, and predict limiting equivalent widths that deviate from the empirical behavior by \leq 5% when the wavelength and Doppler parameter are in the ranges \lambda = 1150-1750 A and b > 10 km/s. We have tested a number of coaddition algorithms and find the noise properties of individual exposures to be closer to the Poissonian ideal than coadded data in all cases. For unresolved absorption lines, limiting equivalent widths for coadded data are 6% larger than limiting equivalent widths derived from individual exposures with the same signal-to-noise. This ratio scales with b-value for resolved absorption lines, with coadded data having a limiting equivalent width that is 25% larger than individual exposures when b \approx 150 km/s.