ANALYSIS AND MODELING OF NONSTATIONARY GROUND MOTION COHERENCY

Author

Habeeb F. Abbas, Southern Illinois University CarbondaleFollow

Date of Award

5-1-2018

Degree Name

Doctor of Philosophy

Department

Engineering Science

First Advisor

Tezcan, Jale

Abstract

Conventional coherency models which rely on Fourier-based tools do not consider the nonstationary character of ground motions. Accurate understanding for nonstationary coherency can motivate accurate modeling and simulation of spatially incoherent ground motions. In this work, a nonstationary ground motion coherency analysis and modeling are performed using wavelet analysis and relevance vector machine regression. To perform the analysis, earthquake ground motion data from four events recorded at dense seismograph SMART-1 array in north-south and east-west horizontal directions are used to investigate the lagged coherency behavior. Continuous Wavelet Transform (CWT) is used to compute the nonstationary lagged coherency that characterizes the space-time variation of seismic ground motion. Complex Morlet wavelet is chosen as the mother wavelet as it is quite well localized in both time and frequency space. Wavelet transform with significance tests have been used to distinguish the strong motions S-wave window with the corresponding frequency band. Based on these S-wave windows, wavelet transform is used to compute the nonstationary lagged coherency of ground motions. A homogeneous isotropic field is assumed. Lagged coherency behavior with frequency, distance, and time is examined. It is shown that the lagged coherency is not constant and evolves with time. The results implied that the lagged coherency on uniform soil depends mainly on time, frequency, and separation distance. The wavelet transform results from the four events are compared with two conventional lagged coherency models. The comparisons showed the difference between the evolving behavior of the lagged coherency with the stationary prediction of the conventional coherency models. To perform the modeling, Relevance Vector Machines (RVM) regression is used to develop a model for the nonstationary lagged coherency that characterizes the space-time variation of seismic ground motion. Earthquake ground motion data from dense seismograph SMART-1 array is used to determine the lagged coherency model in the north-south and the first principal directions. Frequencies and separation distances for the estimated nonstationary lagged coherency represent the trained data for the RVM set to construct the model and estimate the lagged coherency values. The RVM model results are compared with two conventional lagged coherency models. The proposed model does not require a fixed parametric functional form; it can estimate the lagged coherency for different separation distances at different time instants and frequencies. Finally, in this study, the developed nonstationary lagged coherency model is used to generate incoherent artificial spatial ground motions. The simulation of ground motion stochastic process is performed using specified evolutionary power and cross spectral densities. The simulation results are compared with generated ground motion using conventional stationary coherency model. The phase and phase difference from both simulations are compared to show the difference between the results.