Seismic response change of a sand soil after strong motion, revealed by a transmission laboratory experiment

Strong ground motion occasionally reduces the seismic velocity of ground surface layers (Sawazaki et al. 2015) and peak frequencies of the site amplification factor (Sawazaki et al. 2009). Sawazaki et al. (2009) showed a sudden drop in the peak frequency of the ratio of the coda spectral amplitudes observed at surface and subsurface seismic stations after strong ground motion. They also found that the peak frequency had recovered in several days to years and speculated that the recovery process resembles slow dynamics based on a rock experiment. However, their analysis lacked resolution in time because they used coda waves of natural earthquakes.

Laboratory experiments enable continuous monitoring using elastic waves of various amplitudes and frequencies, which may reveal the continuous recovery process. Given a narrowband analysis, TenCate et al. (2000) observed the drop and recovery of the fundamental resonance frequency of an elongated cylindrical rock sample (50 mm × 300 mm) after imposing a strong vibration. Then, the next question arises: how does the material property change besides the fundamental resonance frequency?

In this study, we experimented using a wide-ranged sweep signal (0-20 kHz) for continuously monitoring the change of elastic properties of a soil layer containing dimethylpolysiloxane after adding strong sine signal (0.1kHz). A shaker inputs the sweep and sine signal into the sand layer via a mortar layer; the sand and mortar layers mimic a sedimentary layer and a bedrock, respectively. Continuous sweep waves and sine waves were observed by accelerometers embedded in the sand layer (Fig.1).

We confirmed that the amplitudes decreased in 6, 9, and 12 kHz and increased in 11 kHz just after the strong motion, and both changes recovered gradually (Fig.2). Even without the coda, our result is consistent with that of Sawazaki et al. (2009). We also confirmed that the deeper position tends to show the more significant change and recovery of slowness that had been decreased in 6,9,12 kHz and increased in 5,8kHz. The result of velocity change and recovery was similar to Sawazaki et al. (2015).