Two decades of nonlinear soil response through velocity change analysis in Iwate Prefecture, Japan

It is widely known that the local geology can strongly affect the ground motion by modifying the amplification, duration, and spatial variability of the earthquake shaking. In certain cases, when the ground motion is strong enough, the material may develop large deformations, altering the physical properties of the medium, reducing the shear modulus, increasing the damping, producing liquefaction and permanent displacements among other things. These phenomena belong to the domain of nonlinear soil behavior.
In this study, we use earthquake records collected between 2000 and 2022 from KiK-net stations in Iwate Prefecture (Japan). We investigate three signal processing techniques—deconvolution, phase correlation, and phase autocorrelation—on the earthquake data, focusing on their ability to determine empirical Green’s functions. Our findings show that all three methods give consistent results. Additionally, we group empirical Green’s functions by Peak Ground Acceleration (PGA) into seven bins from 1 to 400 cm/s² and compute an average for each bin. We then apply the stretching technique to determine the velocity change, using the 1-5 cm/s² PGA bin as a reference. This low PGA level is supposed to have linear behavior. We observe that velocity changes increase with increasing PGA. The percentage of velocity changes differs among stations, showing site-specific variations that are not directly correlated with the conventional soil classification based on V_S30.
We also investigate temporal variations of velocity changes at each station. We observe a drop in velocity after strong earthquakes, followed by a long-term recovery. This study proposes a new approach to investigate spatial and temporal, linear and nonlinear soil response.