Surface wave monitoring using ambient noise for detecting temporal variations in underground structures in landslide area

Abstract

The temporal variation in subsurface structure of a landslide area was monitored using spatial autocorrelation (SPAC) as a simple and robust seismic observational method. The SPAC method is often used in civil engineering to estimate one-dimensional (1D) subsurface structures. However, in this study, we monitor the temporal variation in the SPAC coefficient deviation to evaluate environmental change effects and changes in the subsurface velocity structure. We used a seismic array comprising 10 seismometers in a landslide-prone area in Morimachi Town, western Shizuoka Prefecture, Japan, and continuous observations were performed from October 2020 – May 2022. We obtained a 1D velocity structural model as a reference using the multi-mode SPAC (MMSPAC) method with the averaged SPAC coefficients at different distances for all observation periods. We calculated the daily variation in SPAC coefficient relative to the average for all observation periods. We then applied cluster analysis to the SPAC coefficient deviation, which revealed weekly changes likely due to human activity and the effect of stream surges on nearby streams. After removing stream surge clusters and correcting for the weekly change, we reapplied cluster analysis to identify two major clusters. The differences between the two clusters can be attributed to structural changes in the very near surface (∼5 m) and deeper parts (∼20 m), likely influenced by shallow groundwater due to rainfall. By investigating the location of the landslide mass near the observation site, we suggest that structural changes around 20 m deep may correspond to the depth of potential slip surfaces.