What a Binary Black-Hole Signal Measures

Scrollable formula. Use Left and Right arrow keys to reveal hidden content.

Scrollable table. Use Left and Right arrow keys to reveal additional columns.

Scrollable figure. Use Left and Right arrow keys to inspect hidden content.

A binary-black-hole signal directly constrains a detector-frame waveform: its phase evolution, relative amplitudes and arrival times across detectors, and consistency with noise. Masses, spins, distance, sky position, and remnant properties are parameters of a waveform model fitted to those data. Source-frame masses, redshift, and formation history require additional cosmological or population assumptions. Keeping this ladder explicit prevents a posterior label from being mistaken for a raw observable.

Helpful background. Binary Parameters and Observables and Parameter Degeneracies develop the event likelihood. Hierarchical Population Likelihoods explains what changes when events are combined.

The detector-facing signal

For detector $I$ ,

d_I(t)=n_I(t)+F_I^+(t;\boldsymbol n,\psi)h_+(t-\tau_I) +F_I^\times(t;\boldsymbol n,\psi)h_\times(t-\tau_I).

The calibrated data $d_I$ , noise model, detector responses $F_I^A$ , and delays $\tau_I$ define the measurement. Under a waveform family $h(\theta)$ and a stationary-Gaussian residual approximation, the likelihood is controlled by

\ln \mathcal L(\theta) =-\frac12\sum_I \bigl(d_I-h_I(\theta)\mid d_I-h_I(\theta)\bigr)_I+C.

Every inferred source parameter is conditional on the calibration, PSD, analyzed segment, glitch treatment, waveform family, prior measure, and detector network. A high signal-to-noise ratio narrows some likelihood directions; it does not remove model dependence.

Which combinations are well measured

Inspiral phase measures the detector-frame chirp mass

\mathcal M^{\rm det} =\frac{(m_1^{\rm det}m_2^{\rm det})^{3/5}} {(m_1^{\rm det}+m_2^{\rm det})^{1/5}}

more sharply than the individual masses. Higher-order phasing constrains mass ratio in correlation with aligned spin. The effective spin

\chi_{\rm eff} =\frac{m_1\boldsymbol\chi_1\!\cdot\!\hat{\boldsymbol L} +m_2\boldsymbol\chi_2\!\cdot\!\hat{\boldsymbol L}} {m_1+m_2}

is usually better measured than either spin magnitude or tilt. Precession and higher harmonics can break degeneracies, but only when the signal orientation, mass ratio, bandwidth, and signal strength make them visible.

Amplitude scales approximately as $D_L^{-1}$ and is modulated by inclination and antenna response. Consequently, luminosity distance and inclination are correlated. Arrival-time differences localize the source to network-delay surfaces; amplitude and polarization consistency further restrict them. A two-detector event can be loud yet poorly localized or polarization degenerate.

The merger and ringdown constrain detector-frame remnant mass and spin through the full waveform or a declared postmerger model. Inferring radiated energy or recoil then uses numerical-relativity fits and their calibration domains. These derived quantities are not independent measurements.

Source frame is an inferred layer

Cosmological propagation gives

m_a^{\rm det}=(1+z)m_a^{\rm src}, \qquad D_L=D_L(z;\mathcal C),

where $\mathcal C$ is a cosmological model. Without a host redshift, the data do not separately measure $m^{\rm src}$ and $z$ . A cosmology maps the distance posterior to redshift; a population mass model can further sharpen the mapping. Thus “a black hole of source-frame mass $m$ ” carries a cosmology and sometimes a population prior.

Classification is another inferred layer. A compact object in a particular mass interval is not identified as a black hole solely by that mass. The label can use tidal constraints, an assumed maximum neutron-star mass, absence of matter signatures, and a population model. Binary-black-hole interpretations are often strongly supported, but their assumptions should be named.

Current observational interface

At the evidence cutoff 2026-08-18, GWTC-5.0 is the official cumulative catalog baseline. Its observations paper reports candidate selection and event-level source-property products, while the methods paper specifies pipelines, waveform sets, priors, data-quality treatment, calibration, and release products LVK 2026, observations §§2–3 and LVK 2026, methods §§2–5.

The catalog contains many binary-black-hole candidates and several unusually informative signals, but the durable lesson is not a mutable count. It is that event products are versioned: a candidate threshold, posterior release, waveform update, or calibration correction can change which events enter an analysis and how their parameters are summarized.

Reproduction and systematics

A bounded event reproduction fixes the GWOSC event version, detectors, GPS interval, sample rate, data-quality mask, calibration release, PSD interval and estimator, glitch handling, waveform version, likelihood settings, priors, sampler version, and random seeds. The primary outputs are likelihood or posterior comparisons and residual diagnostics—not agreement of one plotted median.

Three independent checks are especially valuable:

recover a simulated signal with known detector-frame parameters and verify credible-region calibration over repeated noise realizations;
compare waveform families that change precession, higher modes, or merger calibration in the regime relevant to the claim;
whiten and subtract posterior-predictive waveforms, checking that coherent residuals are consistent with the declared noise model.

Calibration uncertainty, nonstationary noise, PSD estimation, waveform error, and prior boundaries should be propagated or shown negligible. Source-frame cosmology and population assumptions should be varied only at the layer where they enter.

Exercises

1. Same detector-frame mass, different source frame. A component has $m^{\rm det}=60M_\odot$ . Compare the source-frame masses under $z=0.2$ and $z=1$ .

Solution

Using $m^{\rm src}=m^{\rm det}/(1+z)$ gives $50M_\odot$ at $z=0.2$ and $30M_\odot$ at $z=1$ . The phase can determine the detector-frame mass without selecting between those redshifts; distance, cosmology, and possibly population information do that.

2. Loudness and localization. Why does doubling signal amplitude in the same two nearly aligned detectors not automatically determine both polarizations?