Infrasound signal duration: the effects of propagation distance and waveguide structure

The duration of infrasonic signals recorded by sensors on the Earth’s surface increases with source-to-receiver distance, due to the increased separation of the fastest and slowest portions of the wavefield propagating within atmospheric acoustic waveguides. An algorithm has been developed to consistently measure signal duration, where signals are identified from a combination of multichannel coherence (semblance) and pressure amplitude. The method allows for changes in the signal propagation vector (backazimuth, apparent velocity) across the extent of the signal, which is important for waveguide ducted signals that exhibit multiple arrivals. The algorithm has been applied to 43 infrasound signals exhibiting envelope signal-to-noise ratios ≥3 in the 0.32 to 1.28 Hz passband. All signals were recorded on microbarometer arrays of the International Monitoring System, which is one element of the Comprehensive Nuclear-Test-Ban Treaty verification regime. The data set contains recordings at distances of between 20 and 6300 km from ground truth events that are known to have short source-time functions. The results indicate that signal duration is dependent upon both source-to-receiver range and along-path variability in the strength of the stratospheric waveguide, as measured by the ratio of effective sound speeds in the stratosphere and at the ground. For paths with low along-path waveguide variability the signal duration, D (s), exhibits a weak linear relationship with source-to-receiver range, r (km), such that duration can be estimated as D = 0.278r+ 122. The signal durations are reduced when the propagation paths exhibit waveguide strength variability; in the small number of such cases reported in this paper, signals recorded at source-to-receiver ranges of >3000 km exhibit D ∼300 s. Ray racing simulations indicate that horizontal gradients in waveguide strength act to restrict the signal celerities that can be supported by stratospheric waveguide propagation. Current infrasound signal association and location algorithms do not utilize a signal characteristic from which the source-to-receiver range can be directly estimated. Therefore, the incorporation of a signal duration model into such procedures may provide reduced rates of false signal association and improved event localization.