Bayesian modelling of scattered light in the LIGO interferometers

TL;DR

This paper introduces a Bayesian model for scattered light noise in LIGO data, enabling effective glitch identification and subtraction to improve gravitational wave signal analysis.

Contribution

It develops a physically motivated Bayesian framework for modeling and removing scattered light glitches in LIGO data, enhancing data quality.

Findings

Effective inference of glitch parameters

Successful subtraction of scattered light noise

Improved discrimination of glitch features

Abstract

Excess noise from scattered light poses a persistent challenge in the analysis of data from gravitational wave detectors such as LIGO. We integrate a physically motivated model for the behavior of these "glitches" into a standard Bayesian analysis pipeline used in gravitational wave science. This allows for the inference of the free parameters in this model, and subtraction of these models to produce glitch-free versions of the data. We show that this inference is an effective discriminator of the presence of the features of these glitches, even when those features may not be discernible in standard visualizations of the data.