Unveiling Cryosphere Dynamics by Distributed Acoustic Sensing and Data-Driven Hydro-Thermo Coupled Simulation

Abstract

As global warming continues, the Earth's cryosphere is experiencing severe degradation. This study leverages a novel combination of distributed acoustic sensing (DAS) and artificial intelligence to monitor and decipher cryospheric dynamics. We have developed an advanced time-lapse surface wave analysis workflow to capture shear wave velocity changes $(\mathrm{\Delta }v)$ during a 2-month controlled permafrost thaw experiment in Fairbanks, Alaska. To understand the underlying physical mechanisms of $\mathrm{\Delta }v$, multimodal rock-physics simulations were conducted to associate the observed $\mathrm{\Delta }v$ to hydrological and thermal processes like heating and rainfall events. Furthermore, we employ a physics-guided deep learning algorithm alongside interpretable techniques to evaluate the impact of various physical factors and shed light on the cryospheric hydro-thermo coupling mechanisms. This study highlights the potential of using DAS and data-driven rock-physics simulation for complex cryosphere monitoring and offers a comprehensive view of the permafrost's thawing dynamics.