Nonlinear Soil Behavior During Recent Strong Earthquakes

As shown above, for a detailed study of soil behavior in strong ground motion, records from in situ observations are needed. There still remain significant differences between the conditions of in situ dynamic loading of soils and laboratory and field conditions. Any information on the in situ soil behavior is extremely important; especially, observations with seismic vertical arrays. Since Seed and Idriss (1970) directly observed nonlinear soil response in a borehole during the 1965 Seattle earthquake, the borehole observations have continually evolved to the point where we see entire networks of borehole instruments, such as, KiK-net in Japan, DART in Taipei basin, SCEC in Los Angeles basin, Bay Bridges in San Francisco and others. A seismic vertical array usually consists of 2–4 three-component accelerometers installed in a borehole on the surface and at various depths, down to ~ 100–200 m. As noted by Archuleta and Steidl (2001), earthquake records obtained by seismic vertical arrays provide us answers to many important questions: Where in the soil column did the response become nonlinear? At what level of strain did the nonlinear response occur? How is it manifested in the observed ground motion? What soil properties that can be measured in situ are most important in affecting nonlinearity? Does the soil return to a state with the same linear or nearly linear response? In what time period? etc. (Archuleta and Steidl 2001). Strong earthquakes of last decades, such as the 1985 Michoacan earthquake, 1989 Loma Prieta earthquake, 1994 Northridge earthquake, 1995 Kobe earthquake, 2000 Tottori earthquake, 1999 Chi-Chi earthquake and others provided new experimental data on soil behavior in strong ground motion and on soil liquefaction. At the same time, they caused new discussions about manifestations of nonlinearity of soil response, ranges of applicability of linear models, features of plastic behavior of clays, etc. (Lomnitz et al. 1995; Aguirre and Irikura 1997; Field et al. 1997; O’Connel 1999, etc.). The strong-motion database is growing quickly, but (as shown in Chap. 2 ) the interpretation of the obtained records is sometimes difficult, and new approaches for data processing are needed to extract from the records more information about soil behavior. This chapter describes a method developed by the author in collaboration with Professor Kojiro Irikura (DPRI, Kyoto University, Japan) for estimating stresses and strains induced in soil layers by strong motion, based on vertical array records. The method was used to construct models of soil behavior during recent strong earthquakes: the 1995 Kobe earthquake, 2000 Tottori earthquake, 1999 Chi-Chi (Taiwan) earthquake, 2003 Tokachi-oki earthquake, and 2011 Tohoku earthquake. Models of soil behavior, i.e., vertical distributions of stresses and strains induced in soil layers by the earthquake, allow us to trace changes in the behavior of separate layers during strong motion and evaluate changes of the shear moduli. The estimates are based on real records, so they are free from assumptions about the features of soil behavior in conditions of strong motion.