Analysis Of Strong Motion Recordings And Its Application Form Wenchuan Earthquake

Field investigation, strong motion observation, model testing and numerical simulation are major research methodologies in earthquake engineering. For great earthquake in concerning area is infrequent, the later two methods are dominated in most time, especially in China, which suffered from many great earthquake, but lacking strong motion recordings before Wenchuan earthquake. The Ms8.0 Wenchuan earthquake in 2008 caused strong ground shake in large area and resulted in tremendous economic loss and more than 80 thousand people injured or killed. Thousands of strong ground motions were recorded in near-fault area during the main shock and dense field investigations were conducted in damaged area. Based on the recordings, the baseline correction, characteristics of velocity and displacement and ground surface tilt in near-fault area and source process were studied. The distribution and comparison of buildings damage data were analyzed and the relationship between buildings damage and strong motion parameters were also studied in this dissertation.1. The history of baseline correction for near-fault strong motion recordings in analog and digital recording phases was detailed systematically revised. The influence of six noise resources was analyzed. Correction methods were divided into two categories and five models and the essentials of each method were analyzed. A test criterion for corrected results was proposed. Based on the complexity, the drifting phenomenon in Wenchuan earthquake was divided in six types, including two unreported types. A better data selection for zero-line adjustment was put forward. Common single segment and two segments correction methods were modified and a new two-phase method was proposed to correct the recordings. The changes caused by correction in peak acceleration, Fourier spectrum and response spectrum were analyzed. Based the corrected results, the period of velocity pulse and its single or double sides characteristics were studied. The distribution area of pulse amplitude more than 14cm/s was displayed. The influence of hanging/foot wall and directivity effects to analyzed. The PGV/PGA was recommended to identify velocity pulse. Wavelet transform was employed to study the different dominant frequency in acceleration for velocity and displacement. Combined with GPS results, the displacement field and the attenuation of permanent displacement were analyzed. The PGA, PGV, PGD and RGD (residual ground displacement) were compared to give preliminary judgment. The great effect of filtering on PGV, PGD and RGD was demonstrated2. According to the dynamic equation of sensor pendulum, discrepancy of influence on three components of acceleration imposed by ground surface tilt was analyzed. Results demonstrate that the ground surface tilt may have obvious influence on horizontal components, especially on the low-frequency band. The estimate method of ground surface tilt by acceleration based on Fourier spectrum ratio and wavelet transform were introduced. The ground surface tilt of near-fault 36 stations during main shock and 12 stations from 6 postshocks by spectrum ratio method were calculated. Main shock results demonstrated that the tilt was mainly less than 1.0°and residual tilt existed in some stations, and the tilt larger than 0.1°was within 30km to rupture trace, and the tilt of 0.01°rarely happened where the rupture distance was more than 100km or the root-mean-square of accerleration is less than 200cm/ s2. The characteristics of tilt of two station nearest to central and front rupture trace, and tilt of normal and parallel to fault, tilt in hanging wall/footwall, tilt in thrust and strike-slip segment, and relationship between spectrum ration were discussed. The results of postshocks showed that the tilt larger than 0.001°was seldom observed, and hardly in Ms≤5.3 earthquakes. Furthermore, the relevance of integration drifting and ground surface tilt of horizontal components was investigated, which showed that the former had a predominant influence on the accelerogram: the integration result of low-frequency stage affected by the ground surface tilt coincided with that of whole accelerogram. But the tilt is not as common as drifting. The discrepancy of Fourier spectrum ratio of uncorrected and corrected accelerogram and optimum value of characteristic frequency were discussed.3. The history of seismic source process inversion using seismic recordings was reviewed and discrepancy using far-field and near-fault recordings was discussed. Based on the systematic description of Seismic source theory, model discretization, mathematic equation assembling, constraint, inversion theory and multiple-time-window inversion method were systematic stated to conduct rupture process inversion using near-fault strong recordings. A method based on condition number to determine optimal constraint factor was put forward and testified by numerical examples. Combing with the tectonic background, surface rupture investigation and existing inversion results, a single-plane rupture model was set up. An appropriate occurrence time was chosen as the origin time, and first time of 44 stations within 120km to rupture trace was picked and corrected. The fault plane was divided into 512 sub-faults with the size of 5km*5km. According to the characteristics of noise in the record and requirements of resolution, the velocity was filtered to 0.05Hz～1Hz to be used in inversion. The theoretical seismograms were calculated by software package COMPSYN, which was modified in favor of formation of inversion equation. The fault slips under 1～4 time windows with different constraint factors and data were calculated using nonnegative least squares. The results fitted well field investigation and other inversion results.4. Seismic damage evaluation development in China was reviewed, and initiation and application of earthquake damage index were introduced. Its development in industrial and national standards was compared. Field investigation on 170 spots about 5 building types using damage index in the WenChuan Earthquake were introduced. Influence of fault, geography and soil condition to damage was demonstrated with specific spots. Statistical relation between damage indexes and intensity zones were brought out, which confirmed the classification of buildings in the Chinese Seismic Intensity Scale (2008). Reasons for average damage index exception inⅥandⅪzones were presented. The characteristics of damage index attenuation with fault distance were analyzed and double-log-linear function was proposed to fit their relation1. The fitting parameters were given. Based on the comparison and attenuation characteristic, earthquake resistance capability and correlation of 5 building types were studied. A normalization method was proposed to analyze the relation between PGA and damage index. It showed normalized horizontal PGA was close to brick-concrete building damage index in attenuation and the vertical would underestimate. Pairs of strong motion station and filed investigation spot whose distance was less than 0.1°were chosen to analyze the correlation between earthquake damage index and seismic parameters, including PGA, PGV and response spectrum amplitude. Results showed that the linear correlation coefficient between damage index of brick-concrete building and brick-timber building and PGA, PGV lied in 0.7 to 0.8, but its bellow 0.3 for old buildings, and there was no significant difference between PGA and PGV. The correlation between damage index and response spectrum amplitude was similar to the above two. Generally, it would be more reliable to predict earthquake damage of brick-concrete and brick-timber buildings those strong motion parameters.