Assessment of Arrow-of-Time Metrics for the Characterization of Underwater Explosions

Anthropogenic impulsive sound sources with high intensity are a threat to marine life and it is crucial to keep them under control to preserve the biodiversity of marine ecosystems. Underwater explosions are one of the representatives of these impulsive sound sources, and existing detection techniques are generally based on monitoring the pressure level as well as some frequency-related features. In this paper, we propose a complementary approach to the underwater explosion detection problem through assessing the arrow of time. The arrow of time of the pressure waves coming from underwater explosions conveys information about the complex characteristics of the nonlinear physical processes taking place as a consequence of the explosion to some extent. We present a thorough review of the characterization of arrows of time in time-series, and then provide specific details regarding their applications in passive acoustic monitoring. Visibility graph-based metrics, specifically the direct horizontal visibility graph of the instantaneous phase, have the best performance when assessing the arrow of time in real explosions compared to similar acoustic events of different kinds. The proposed technique has been validated in both simulations and real underwater explosions.