Statistical Models for Infrasonic Propagation: Application to the Detection Capability of the Global IMS Network

Abstract The detection capability of the global infrasound International Monitoring System (IMS) deployed to monitor compliance with the Comprehensive Nuclear‐Test‐Ban Treaty is highly variable in space and time. Previous studies estimated the source energy of near‐surface explosions from remote observations using empirical yield‐scaling relations. However, these relations simplify the complexities of infrasound propagation. In order to reduce the variance in the predicted wave attenuation, massive numerical propagation simulations are carried out by exploring a wide range of realistic atmospheric scenarios. An analytical expression is proposed to model transmission losses at distances up to 4,000 km. This attenuation relation is validated using multi‐year near‐ and far‐field records of volcanic eruptions from Mount Etna, Italy, as a benchmark. An idealized explosive source model is combined with transmission loss and measured noise statistics to quantify the 90% probability detection threshold of the IMS network. This approach yields high‐resolution detection capability simulation results with limited computational resources. Simulations predict that explosions equivalent to 500 t of TNT equivalent would be detected by at least two stations at any time of the year, with the detection capability being best during the solstice periods. Due to the high spatio‐temporal variability of the winds in the ground‐to‐stratosphere levels, threshold variations can reach one order of magnitude from seasonal down to semi‐diurnal scales.