Effect of noise estimation in time-domain ringdown analysis

Accurate noise estimation from gravitational wave GW data is critical for Bayesian inference. However, recent studies on ringdown signal, such as those by Isi et al. [Phys. Rev. Lett. 127, 011103 (2021)], Cotesta et al. [Phys. Rev. Lett. 129, 111102 (2022)], and Isi and Farr [arXiv:2202.02941], have encountered disagreement in noise estimation, leading to inconsistent results. The key discrepancy between these studies lies in the usage of different noise estimation methods, augmented by the usage of different sampling rates. We achieved consistent results across various sampling rates by correctly managing noise estimation, shown in the case study of the GW150914 ringdown signal. By conducting a time-domain Bayesian inference analysis on GW data, starting from the peak of the signal, we discovered that the first overtone mode is weakly supported by the amplitude distribution, with a confidence level of $1.6\sigma$, and is slightly disfavored by the log-Bayes factor. Overall, in our time-domain analysis we conclude there is no strong evidence for overtones in GW150914.