Application of Adaptive Spline Regression and wavelet based Method on Glitches in Gravitational Wave data

Data from the LIGO and the Virgo gravitational wave (GW) detectors contain transient non-GW signals, or “glitches”, that significantly degrade the sensitivities of the search algorithms for the binary inspiral and merger signals (the type observed so far). For about 20% of GW signals in the third LIGO-Virgo observing run, the proximity of or, in some cases, direct overlap with glitches requires estimating and subtracting the latter out from the data. However, this is challenging because of the unpredictability and diversity of glitch waveforms. We address this challenge using an adaptive spline fitting method called SHAPES [Mohanty & Fahnestock, 2019]. Compared to the established wavelet-based WaveShrink [Donoho & Johnstone, 1994] denoising method, SHAPES preserves GW signal power at low frequencies better. For some types of glitches, combining SHAPES and WaveShrink is more beneficial. Our results include the removal of the famous large amplitude glitch coincident with the GW170817 event, an egregious example of the adverse effects of glitches.