Compactness Inference in Gravitational-Wave Mergers with PhenomDECO: Catalog Benchmarks and Robustness Diagnostics
Several gravitational wave (GW) observations have been identified as binary black hole (BBH) mergers, including systems with component masses that challenge typical formation scenarios. These observations motivate broader tests of whether the detected sources are consistent with this interpretation. We address this question using~\deco~, an existing phenomenological extension of a BBH model that uses an effective compactness parameter to characterize departures from the expected merger morphology. Applying this model to all high-significance BBH events from GWTC-3, we establish~\deco~as a robust test of the nature of compact binaries. In preliminary analyses we identify three recurring posterior morphologies: (i) near-Gaussian peaks consistent with the BBH expectation $C\sim0.5$, seen in 60\% of events; (ii) posteriors with additional high-compactness support $(C\ge0.8)$; and (iii) dominant low-compactness modes near $C\sim0.15$ in $\sim 20\%$ of cases. For the latter, the low-compactness modes disappear when the data, especially from Livingston, are analyzed from a higher starting frequency, indicating sensitivity to low-frequency noise artefacts. We further use time--frequency residuals, computed after subtracting maximum-likelihood BBH and~\deco~waveforms from the strain data, to assess if the data is better described by a compactness-based deformation. With this analysis, we conclude that all of the GWTC-3 observations that we have considered are indeed consistent with BBH sources. The exception is the high-mass GW231123 signal, for which data from \emph{both} detectors must be analyzed above 50Hz to remove a low-compactness mode. This study shows that low-frequency data treatment is crucial before attributing apparent deviations from BBH expectations to exotic physics, and provides a benchmark for compactness-based tests of merger morphology in current and future GW detections.