Influence of rainfall in a short-term seismic velocity monitoring at an urban landslide site in Minas Gerais, Brazil

Abstract

Most high-magnitude landslides in Brazil are triggered by rainfall. The partial or total liquefaction of the soil mass is directly related to an increase in the pore pressure, which can happen after or during heavy and sustained rainfalls. Traditional near-surface geophysical methods for saturation and rigidity assessment usually involve short-duration data acquisition campaigns, reporting results representative of a specific moment. On the other hand, the cross-correlation of the ambient noise wavefield registered at different locations in continuous data streams has emerged as a promising alternative for monitoring seismic velocity variations in time. In the present study, we evaluate if the observed seismic velocity perturbations can satisfactorily relate to rainfall periods on a local scale and within minutes of resolution at a slope in Itabirito, state of Minas Gerais, Brazil, where landslide activity occurred. We calculate velocity changes in the tail portion (coda) from Green's functions retrieved from cross-correlation of seismic noise time series registered at pairs of stations from September 2022 to November 2022. In the 50 days observation period, the monitored area remained stable. We identified clear time intervals where the relative seismic velocity decreases after or during rainfall, which indicates modification in the saturation conditions of the soil. The results demonstrated that seismic velocity changes obtained from the coda of cross correlations of the ambient noise wavefield could satisfactorily constrain rainfall periods and indicate possible changes in the saturation conditions of the soil at a small local scale with a high temporal resolution.