An automated pipeline for scattered light noise characterization

Scattered light noise (or scattering) affects the sensitivity of gravitational wave detectors in their detection frequency band. The mitigation of such nonlinear and nonstationary noise can be carried out experimentally and employing data analysis techniques, e.g., applying adaptive algorithms to the data affected by noise. We present gwas , a fully automated pipeline based on the time-varying filter empirical mode decomposition (tvf-EMD) algorithm, to identify, characterize, and monitor objects inducing scattering to the gravitational wave detector's output. The tvf-EMD algorithm is suitable for decomposing signals with time-dependent frequency, such as scattering. The pipeline application to LIGO Livingston data shows that most of the scattering noise present in the third observation run was due to the penultimate mass at the end of the X-arm of the detector (EXPUM), whose motion is excited in the 0.1 Hz to 0.3 Hz frequency range (so-called microseismic band). Furthermore, we show that the pipeline can perform daily analyses, and we test it on six days of LIGO Livingston data. These analyses allowed monitoring of the onset and time evolution of scattering due to the EXPUM in connection with the variability of microseismic band noise measured at the detector site.