MODELLING OF STRONG MOTION GENERATION AREA USING SEMI EMPIRICAL TECHNIQUE

Abstract

Strong motion studies play a significant role in seismology and earthquake engineering. A fundamental problem for civil engineers is to design structures that can withstand any probable earthquake in the region during its lifetime. Designing of such structures requires reliable prediction of strong motion parameters. These parameters can be obtained either directly from past observed records or from simulated strong motion records. In a region with limited information of recorded seismicity simulation of strong motion is the only option left for estimation of the safe engineering design parameters. It also provides an efficient way to study various property of the earthquake source. This has motivated a need for a technique of simulation of strong ground motion which should depend on the easily available parameters at any new site. In the recent years the method of modified semi empirical simulation of strong ground motion has evolved as an effective tool to simulate strong motion records. The semi empirical method has been extensively tested for its applicability in simulation of strong ground motion by Midorikawa (1993), Joshi and Patel (1997), Joshi et al. (1999, 2001, 2012a, b), Kumar et al. (1997), Joshi (2001, 2004), Joshi and Midorikawa (2004), Joshi and Mohan (2008). In the present thesis modifications in the semi empirical technique has been introduced to incorporate the effect of SMGAs in rupture model. In all earlier developments in the semi empirical approach (Joshi et al. 2012a, b, 2014) frequency independent radiation pattern has been used. In order to simulate strong ground motion more accurately, the semi empirical method has been further modified to incorporate frequency dependent radiation pattern in the present thesis. To validate the present modified technique strong ground motions from the Tohoku earthquake (Mw 9.0) of 2011, the Niigata earthquake (Mw 6.6) of 2007 and the Uttarkashi earthquake (Mw 6.8) of 1991, respectively has been simulated. To validate the developed technique, synthetic ground motions have been simulated for the Tohoku earthquake. Strong motion generation areas identified within the rupture plane of the Tohoku earthquake have been modeled using present modified technique. Two different source models having four and five SMGAs respectively are considered for modelling purpose. Strong motion records using modified semi empirical technique have been simulated at two near field stations located at epicentral distance of 137 km and 140 km, respectively. Comparison of the ii observed and simulated acceleration waveforms is made in terms of root mean square error (RMSE) at both stations. Minimum root mean square error of the waveform comparison has been obtained at both the stations for source model having five SMGAs. Strong ground motion have been simulated from same rupture at twelve stations lying at epicentral distance between 137 km and 329 km, respectively. Comparison of observed and simulated records has been made in terms of RMSE in acceleration records and response spectra at twelve stations. Simulations have been made at twelve stations to obtain distribution of peak ground acceleration and peak ground velocity with epicentral distance. Peak ground acceleration and response spectra from simulated and observed records are compared at twelve stations surrounding the source of the Tohoku earthquake. Comparison of waveforms and parameters extracted from observed and simulated strong motion records confirm the efficacy of the developed modified technique to model earthquake characterized by SMGAs. Strong ground motions for the Tohoku earthquake has been also simulated using frequency dependent and independent radiation pattern. A frequency dependent radiation pattern is used to simulate high frequency ground motion more precisely. Identified SMGAs (Kurahashi and Irikura 2011) of the Tohoku earthquakewere modeled using this modified technique. We analyzed the effect of changing seismic moment values of SMGAs on the simulated acceleration time series. Final selection of the moment values of SMGAs is based on the root mean square error of waveform comparison. Records are simulated for both frequency dependent and constant radiation pattern function. Simulated records for both cases are compared with observed records in terms of peak ground acceleration and response spectra at different stations. Comparison of simulated and observed records shows that frequency dependent radiation pattern function reduces the RMSE. The results confirm the efficacy of the developed modified technique. Modified technique has been used to simulate strong motion records of the Niigata earthquake of 16th July, 2007 of magnitude 6.6 (Mw). Source model of this earthquake has been computed from accelerograms recorded by KiK-net at near field stations surrounding source of earthquake. Several isolated wave packets were seen in recorded accelerograms at near field stations surrounding source of this earthquake. Each wave packet in recorded accelerogram represents an isolated patch of envelope of accelerogram released from a rupture plane and is considered to be an independent source of strong motion generation area. Three different isolated wave packets have been identified within the rupture plane of the Niigata earthquake from iii recorded accelerograms. These isolated wave packets were considered as strong motion generation areas (SMGAs) in the rupture plane. Source parameters of each SMGA were calculated from the source displacement spectra. The approximate locations of SMGAs over the source fault were estimated using spatio-temporal variation of 48 aftershocks recorded by KiK-net and K-NET. Modified semi empirical method has been used to simulate strong ground motion at various stations. Comparison of the observed and simulated acceleration waveforms and response spectra is made in terms of root mean square error (RMSE). Comparison of observed and simulated records at eight stations confirms the suitability of final source model consisting of three SMGAs and efficacy of the modified semi empirical technique to simulate strong ground motion. The Uttarkashi earthquake of 20th Oct., 1991 of magnitude 6.8 (Mw) has been used as a case study in the present thesis. The proposed approach of modelling ground motion has been validated by simulating ground motion for the Uttarkashi earthquake of October 20, 1991. In this thesis SMGAs of the Uttarkashi earthquake have been modeled. Two different isolated wave packets in the recorded accelerogram have been identified from recorded ground motion, which accounts for two different SMGAs in the entire rupture plane. The approximate locations of SMGAs within the rupture plane were estimated using spatio-temporal variation of 77 aftershocks. Source parameters of each SMGA were calculated from theoretical and observed source displacement spectra computed from two different wave packets in the record. The final model of rupture plane responsible for the Uttarkashi earthquake consists of two SMGAs and the same has been used to simulate horizontal components of acceleration records and response spectra at different station using modified semi empirical technique. Comparison shows that simulated record bears realistic shape as that of observed record and the peak ground acceleration of observed and simulated record is also comparable. This confirms the efficacy of approach and suitability of the final model to predict strong motion records of the Uttarkashi earthquake. Several strong motion parameters are required for the estimation of seismic hazard in any region which is sometimes not easily available at all sites. This thesis presents modified semi empirical method which has advantage of using easily available parameters. The technique can be applied to a region having scarcity of observed strong motion data and can be effectively used for estimating earthquake resistant design parameters.