Empirical prediction of seismic strong ground motion : contributions to the nonlinear soil behavior analysis and the Empirical Green's function approach

Abstract : Seismic hazard assessments must consider different aspects that are involved in an earthquake process and affect the surface ground motion. Those aspects can be classified into three main kinds. 1) the source effects are related to the rupture process and the release of energy. 2) the path effects related to the propagation of energy inside Earth. 3) the influence of the shallow layers geotechnical characteristics; the so-called site-effects. The site effects are considered in risk mitigation through the evaluation of the seismic soil response. Under cyclic solicitations the soil shows a non-linear behavior, meaning that the response will not only depend on soil parameters but also on seismic motion input characteristics (amplitude, frequency content, duration, …). To estimate the non-linear site response, the usual practice is to use numerical simulations with equivalent linear analysis or truly non-linear time domain approach. In this document, we study the influence of the nonlinear soil behavior on the seismic site response by analyzing the earthquake recordings from borehole array configurations. We use the Kiban Kyoshin network (KiK-Net) data. All 688 sites are instrumented with two 3-components accelerometers, one located at the surface and the another at depth. From these data, we compute the ground motion amplifications from the surface to downhole recordings by the computing Fourier spectral ratios for the aim to compare between the spectral ratio for weak and strong ground motion. The main effect of the non-linear behavior of the soil on the site transfer function is a shift of the amplification towards lower frequencies. We propose a new methodology to quantify those changes and study the nonlinear effects. This work results in a site-dependent relationship between the changes in the site response and the intensity parameter of the ground motion. The method is tested analyzing the records of the earthquake of Kumamoto (Mw 7.1, 2016). Posteriorly, we propose to integrate a correlation between seismic moment and the duration of the fault (Courboulex et al., 2016) in the empirical Green’s function method. This methodology was applied to simulate one seduction event in Equator, and we compare the results with the records of the Pedernales earthquake (Mw 7.8, 2016) in the city of Quito. We attempt to take in account the nonlinear effects in the empirical Green’s function method. We use the methodologies of the first part of this document based on the frequency shift parameter. The procedure could be implemented in other methodologies that can predict an earthquake at a rock reference site, such as the stochastic methods. We test the procedure using the accelerometric records for one of the aftershocks o the Tôhoku earthquake (Mw 7.9).
Document type :
Liste complète des métadonnées

Contributor : Abes Star <>
Submitted on : Monday, April 15, 2019 - 12:03:07 PM
Last modification on : Tuesday, April 16, 2019 - 1:29:30 AM


Version validated by the jury (STAR)


  • HAL Id : tel-02099814, version 1



David Alejandro Castro Cruz. Empirical prediction of seismic strong ground motion : contributions to the nonlinear soil behavior analysis and the Empirical Green's function approach. Earth Sciences. Université Côte d'Azur, 2018. English. ⟨NNT : 2018AZUR4216⟩. ⟨tel-02099814⟩



Record views


Files downloads