Accounting for detector calibration uncertainty in black-hole ringdown analysis

Abstract

Detector calibration systematics can bias measurements of a remnant black hole's mass and spin, producing spurious evidence against general relativity if unaccounted for. The ringdown phase in the gravitational-wave signal produced by a binary black-hole merger encodes the remnant properties: at sufficiently late times, the signal is well described by a superposition of quasinormal modes, each with a frequency and damping time set only by the remnant mass and spin. Here we present two complementary approaches for analyzing the ringdown portion of the data to estimate the remnant mass and spin in the presence of a nonnegligible detector calibration error, using the quasinormal-mode rational filter (QNMRF). With a well-measured error, we apply the best-estimated calibration correction to the strain before conducting the standard analysis; with an unknown error, we instead jointly sample the remnant mass, spin and the frequency-dependent calibration error, marginalizing over an uncertainty prior. By injecting the numerical-relativity waveform SXS:BBH:0305 into colored detector noise, at a full-signal matched-filter signal-to-noise ratio of $\sim60$ and with simulated calibration errors of $\pm25\%$ in magnitude and $\pm25^\circ$ in phase, marginalizing over the calibration uncertainty recovers the true mass and spin within the 90\% credible interval, whereas the estimated values are biased if the calibration errors are ignored. We apply the method to GW250207, an event recorded while the Hanford detector response was not well characterized, and recover a remnant consistent with the inspiral-merger-ringdown prediction using the ringdown portion alone, both by correcting for a best-estimated calibration error and by marginalizing over a broad uncertainty prior.