Temporal change of seismic velocity and site response for different scales and implications for nonlinearity
MetadataShow full item record
This thesis consists of two major parts. In the first part, I monitor the temporal change of S-wave velocity in shallow soil layers using seismic data collected in an experiment at Panola Mt. Atlanta, GA, 2006. I use the cross correlation function to find the arrival time differences for different water levels, and then calculate the change of Rayleigh wave phase velocity according to different frequencies in the range 5 to 50 Hz. After that, I find a reference 1-D layered P and S-wave velocity model from the measured Rayleigh wave dispersion curve, and put 6 sets of Gaussian perturbations into the reference velocity structure to invert for the actual temporal change of velocity structure in the experiment. I find a clear increase of S-wave velocity in the water injection area, and the S-wave velocity gradually recovers to the initial value after we stop pumping water. In the second part, I analyze temporal changes in fault zone site response along the Karadere-Düzce branch of the North Anatolian Fault, starting 8 days before and ending 72 days after the 1999 Mw7.1 Düzce, Turkey, earthquake. The analysis involves comparisons of strong motion seismic records at station VO inside the Karadere fault and station FP about 300 m away from the fault. I compare all available seismic waveforms at these stations, including those generated by foreshocks, the mainshock, aftershocks and seismic noise, and cut them into 10 s windows with a 5 s overlap. Fourier amplitude spectra are computed for seismic data in each window, and the average amplitude spectra for the two horizontal components are used to obtain the spectral ratio for each on/off fault pair of seismic records. The spectral ratios are smoothed over every 10 points in the frequency domain (0.5 Hz). The results show a shift of the spectral peak to lower frequencies during the main shock. The peak frequency reduces from 4.3 Hz several days before the main shock to 2.9 Hz (67.4% of the pre mainshock value) right after the mainshock. It quickly recovers to 3.8 Hz (64% recovery of the dropped value) after a day, and then gradually recovers to 4.0 Hz (79% recovery of the dropped value) after 72 days. I also compare the results from all the seismic data including direct S-wave, S coda waves and seismic noise and from coda waves only and find that the results from coda waves which are generally less scattered than those from all the data, and show lower amplitude of spectra ratio with higher peak frequencies. The observations suggest a nonlinear behavior of the fault zone material under strong ground motion of nearby major earthquakes. Finally I attempt to link the two parts by identifying their implications for the nonlinear site effects.