Single and multiple sensor identification of avalanche-generated infrasound

Abstract

The ability to identify snow avalanches as they occur is essential for aggressive avalanche management in transportation corridors and is a fundamental ingredient of avalanche forecasting. Past studies have shown that moving avalanches emit a detectable sub-audible sound signature in the low frequency infrasonic spectrum. Experimental infrasound avalanche monitoring activities conducted in the United States Rocky Mountain West clarify avalanche event identification capabilities of single sensor and multiple sensor systems. Avalanche identification performance of single sensor monitoring systems vary according to ambient noise and signal levels. While single sensor auto correlation signal processing algorithms identify avalanche activity, uncertainties (i.e. false negative identifications and false positive identifications) increase as wind noise increases, and as signal levels decrease due to increasing distance or smaller sources. Monitoring with multiple sensor systems substantially improves avalanche event identification robustness under windy and noisy conditions, while also allowing location estimates of avalanche events to be made. Avalanche event identification and localization capabilities of cross correlation and semblance multiple sensor signal processing algorithms are demonstrated via a sensor array monitoring system. Also demonstrated are avalanche identification and localization capabilities of distributed networks of infrasound monitoring systems. Garnered knowledge is being ported into near real-time prototype systems that will be operated in the Jackson Hole, Wyoming region. Prototype operation will provide performance evaluations in practical highway and recreational area settings. Reliable implementation of infrasound monitoring technology to automatically identify avalanche events requires further innovative solutions to problematic ambient wind noise and interfering signals.