Volcanic eruptions emit sound waves that are typically dominant at infrasonic frequencies (~0.01–20 Hz) and that have been used to estimate eruption source parameters valuable for hazard mitigation. However, the accuracy of these estimates depends on the ability to recover the pressure-time history of the acoustic source, which may be distorted during propagation even at local recording distances (<15 km). We aim to quantify potential distortions caused by diffraction over topography and wave steepening during nonlinear propagation.

To investigate the effects of topographic diffraction, we evaluate the ability of a thin screen approximation to predict amplitude losses over topography at Sakurajima Volcano, Japan. Using synthetic data from numerical modeling, we show that amplitude losses from diffraction over volcano topography are systematically less than predicted for a thin screen. We propose that attenuation by diffraction may be counteracted by acoustic focusing (constructive interference between reflections along concave slopes). We conclude that thin screens are inappropriate proxies for volcano topography, and maintain that numerical simulations are required to account for wavefield interactions with topography.

To investigate the role of near-source nonlinear propagation in volcano infrasound, we apply a previously developed, quadspectral density-based nonlinearity indicator to observed and synthetic signals corresponding to explosive eruptions at Sakurajima Volcano, Japan and Yasur Volcano, Vanuatu. We hypothesize that significant nonlinearity will be expressed as energy transfer from low to high frequencies as the acoustic waves steepen towards shock waves. At Sakurajima Volcano we find evidence for spectral energy transfer in the synthetic data but inconclusive results from the observed signals, suggesting that nonlinearity signatures may be present but obscured by complicating factors (e.g., topography, wind, waveform undersampling). At Yasur Volcano we find evidence for nonlinearity in both synthetics and observations, suggesting that nonlinearity is better observed at short source-receiver distances and with higher sampling rates. At both volcanoes we estimate that cumulative spectral changes by nonlinear propagation are small (<1% of source levels). We conclude that, for signal amplitudes associated with low-level explosions common in field campaigns, nonlinear propagation does not introduce significant errors to acoustically-based source parameter estimates when compared to a linear assumption.