Monitoring Water Content Variations From Seismic Noise in a Controlled Laboratory Experiment: A Quantitative Approach Combining Poroelastic Predictions With Kernel Wave Sensitivity Weighting

Abstract

Despite the existence of various hydrological and geophysical methods for characterizing the vadose zone and groundwater, it remains challenging to implement cost-effective, accurate, and efficient techniques for their long-term monitoring with high spatial resolution. A growing number of recent studies suggest that seismological methods based on continuous seismic noise recording can potentially address these difficulties. This study presents an original laboratory experiment aimed at assessing the sensitivity of passive seismic interferometry imaging (PII) to controlled fluctuations in water content. To achieve this, we used the recording of the seismic noise generated by a continuous seismic source to reconstruct ballistic surface Rayleigh waves propagating in the [200–500] Hz range within a 1-m scale sandbox. Multiple controlled cycles of water imbibition and drainage at the base of the sandbox produce significant variations in the seismic wavefield and especially in dominant surface waves. The large relative velocity variations $\delta v/v$ (−35%), measured in Rayleigh waves with a fine temporal resolution, match the water pressure measurements conducted within the sandbox. The observations are well predicted by an original theoretical approach combining a Biot-Gassmann-Wood poroelastic model that incorporates effective pressure fluctuations and the frequency-dependent sensitivity kernels of Rayleigh waves. These results confirm the potential of the PII method in monitoring saturation changes in the vadose zone as well as the substantial effect of effective pressure fluctuations, at least when Rayleigh waves dominate ballistic arrivals.