Answers are grounded in this paper and cite their sources.
SUMMARY A method is investigated using density-based clustering on spectrogram data to identify signatures of interest in a single infrasound waveform data stream. In most applications, coherence, array-based analysis is used to distinguish infrasonic pressure fluctuations from those produced by non-acoustic phenomena such as turbulence and ambient wind flow. The dynamic nature of such phenomena produces significant variability in the acoustic background for infrasound analysis and the highly dispersive nature of infrasound propagation results in limited applicability of pattern matching approaches. An approach is investigated here to use a data-adaptive threshold to identify those spectrogram points which are above the background. Density-based clustering is applied to the resulting collection of time–frequency points to identify signatures that may be of interest. The method is applied to infrasonic signals produced by a large explosion in the southeastern United States that occurred in October of 2012. A large number of microbarometers were deployed as part of the USArray; however, the spacing between network nodes (approximately 70 km) is too large to enable coherence analysis. Detections are identified using the spectrogram, density-based clustering approach and comparison with analyst reviewed picks shows a recall rate greater than 98 per cent. The method also identifies additional signatures on several stations across the eastern United States that were not picked by the analyst but can be attributed to the event by comparison with observed and predicted celerity-range trends.
No takes yet. Share an insight, caveat, or question.
Blom et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: