Experimentally observed effects of anomalously weak and strong amplification of seismic waves in soft subsurface soils are studied using the records of KiK-net seismic vertical arrays (of strong motion network of Japan). The amplification was estimated based on peak ground accelerations recorded during ~20 earthquakes at each station. To interpret the obtained results, we engaged models of soil behavior at these sites during the Tohoku earthquake on March 11, 2011.

With this work, we start studying of the response of soils of various types (among 11 types of the most representative surface soils on the territory of Russian Federation) in seismic motion, based on in situ observations (KiK-net data). The response of soft sandy and loess soils to weak and medium-intensity seismic motion is analyzed. We show that the response of rather homogeneous soil profiles (without clear seismic boundaries) to seismic motion is frequency-dependent and contains resonant frequencies, at which seismic motion is noticeably enhanced. At stronger seismic motion, the resonant frequencies are reduced, and the amplification of seismic motion by soil layers is also reduced. These effects are not taken into account in seismic rigidity method, the application of which is still required by the Building codes acting in Russia. We should change the practice of application of the seismic rigidity method in the Building codes, by clearly defining the scope of the method, and we should move to the spectral description of the corrections for ground conditions in earthquake resistant construction, by means of response spectra, as done in other countries.

The normative formulas of SP 286.1325800.2016 for estimation of peak ground accelerations in earthquakes were checked on records obtained in the near-fault zones of strong earthquakes (MS ~6.9÷7.15) of Japan made by K-NET and KiK-net stations. It was found that the recorded values of peak accelerations have a fairly large scattering even in the selected narrow range of magnitudes, which is obviously due to local site effects (soil response, topographic effects, etc.), as well as to differences in the characteristics of the earthquake sources and the paths of seismic wave propagation. Calculations by the normative formulas produced estimates that are near the lower limit of the recorded peak accelerations, and in the near-source zones, they noticeably underestimate the recorded peak accelerations. In general, the formulas of SP 286.1325800.2016 give us only approximate estimates, and therefore, it is necessary to develop more reliable methods for estimating peak accelerations and other parameters of seismic motion, taking into account soil conditions and regional characteristics, with components of probabilistic analysis.