UC Santa Barbara

Acoustic waves below the audible frequencies (20 Hz) are called infrasound. Generated by a wide variety of sources, these signals can propagate thousands of kilometers, aided by their low attenuation as well as strong winds and high temperatures in the stratosphere and thermosphere that enable them to return to the ground. Volcanic eruptions are among the most powerful sources of infrasound, and the International Monitoring System (IMS) infrasound network, maintained by the Comprehensive Nuclear-Test-Ban Treaty Organization (CTBTO), is capable of monitoring volcanic eruptions around theglobe. However, the spatiotemporal variability of the atmosphere poses a major challenge. In particular, strong horizontal cross-winds affect the observed infrasound arrival back-azimuths, causing source mislocations up to hundreds of kilometers. This dissertation presents an automatic iterative method combining empirical climatologies (HWM14/NRLMSIS2.0) and 3D ray tracing (infraGA) to obtain rapid, robust first-order estimates of the back-azimuth deviations, producing corrections for location and characterization procedures. The results compare well with a more realistic model of the lower atmosphere using Numerical Weather Prediction (NWP) simulations. Potential applications include near real-time volcano monitoring and large-scale reanalysis of infrasound archives. To test this method, we first apply it to the eruptions of Puyehue-Cordón Caulle (2011) and Calbuco (2015) in Chile, using data from IMS stations within 5000 km. The back-azimuth corrections reduce the source mislocation by ∼84% (242 to 38.7 km) and ∼75% (366 to 93.1 km), respectively. Then, we assess the method’s temporal accuracy using data from 2003-2022 from IMS station IS22 (New Caledonia), which has recorded persistent volcanic infrasound from Yasur (∼400 km) and Ambrym (∼700 km) volcanoes (Vanuatu archipelago). We find good agreement with the seasonal variation of the dataset, showing we can provide year-long estimates of the back-azimuth deviation. Finally, we use the method to identify long-range infrasound signals recorded at the Antarctic IMS station IS27, attributable to episodes of moderate persistent eruptive activity at Mount Michael volcano, 1,670 km away on Saunders island (South Sandwich islands). Examining data from 2004-2020 reveals previously hidden candidate eruptive signals, especially from May 2005 to January 2008 and from May 2016 to April 2018.