Some Key Issues On The Strong Motion Observation

The strong motion network is under the rapid development in all of the world, thedevastating earthquakes in Chile, Haiti, New Zealand, Japan and Mexico recently havethe chance to accumulate more and more strong motion records which give an excellentopportunity for earthquake engineering research. However, with the high-quality strongmotion data captured, some abnormal strong motion data also appeared. Some abnormaldata could be explanted by the network management and operation or the defect of theinstruments. Some look well, which just hiding some information we maybe have neverknown before. This requires us updating the strong motion processing methodology toanalyze the data from near-fault and far-fault records.In this paper, the typical free field ground motion observation array, buildingstructure array, and dam array are summarized. It laid the emphasis on the precision ofsome basic requirements on the station layout. Some abnormal waveforms are discussed.The causality of filters and related parameters of ground motion parameters are studiedfor data processing. Some permanent displacement calculation methods are compared,the baseline correction method combined the Hermit interpolation and the flatness ispresented, then, typical residual displacements are calculated in the recent earthquakesground motion compared with GPS station nearby. At last, the velocity pulse recognitiontechnology is carried out with wavelet transform, empirical mode decomposition andfilter. The details of some understanding and achievement of main research are asfollows:1. The layout of strong ground motion array, general building structures and damsarray are summarized, and the basic principle of the strong motion stations are analyzed.Combining the characteristics of earthquake damage in the Wenchuan earthquake, itsuggests some recommendations of structure strong motion observation array, which it isbetter to provide reliable data for studying the dynamic response of structures underseismic load in the future. The purpose of this is just to summarize the relevant technicalrequirements of strong motion array and the application in obtaining some backgroundknowledge for the strong motion observation array design.2. Some weird waveforms of previous earthquake are analyzed, such as thephenomenon of asymmetric waveform, and the likely baseline drift in the accelerationtime history, and the strong motion record quality issue is discussed preliminarily. Wefocuse on the phenomenon of "spike". It could be determined by the statistic ratio of thePGA point and adjacent sampling points in acceleration records by processing theChi-Chi, Wenchuan, New Zealand and Japan earthquake data. The effect of eliminatingthe spike on the time history and response spectrum is presented, processing approachesare provided as well.3. The root cause of noise in analog and digital strong motion records is analyzed. Instruments calibration and noise cutoff frequency are discussed. Taking main shock ofthe Wenchuan records and Butterworth as an example, the use of the non-causal filter,adding zero, cosine transition, cut-off frequency effect on the parameters of the strongmotion records and preserving zero are studied, and four causal filters calculation resultswith different filters parameters are compared. Studies have shown that non-causal filterhas the characteristic of the zero phase by adding same length zeros to the endpoints ofrecord; cosine transition between adding zeros and the original record is necessary;calculation of a Butterworth acausal filter is stable; EMD method can be perfect with itï¼›PGV and PGD is more sensitive to the choice of the high-pass cutoff frequency; it isbetter to select Chebyshev I non-causal filter when the ripple factor is0.2dB; however,Chebyshev II non-causal filter should be chosen when the ripple factor is3dB;acceleration should ensure zero to accommodate the coordination among the data afterfilter.4. The baseline drift types are classified, and the new approach combining Hermitinterpolation and the flatness is suggested, and four other methods, such as Iwan I, Boorev0, Wu, Wang are compared with the selected records from Chi-Chi, Wenchuan, Japanearthquake. The permanent displacement field of Japan is calculated. It shows thatbaseline drift is more complicated, and the experience method is usually applied.However, the new approach suggested in this paper reduces the subjective random tosome extent, which shows a better reasonability and efficiency.5. Based on the Fourier transform, wavelet analysis and empirical modedecomposition theory, the acceleration records of which PGA is more than100gal, areselected from the main shock and aftershocks of Wenchuan earthquake, Mw7.1andMw6.3earthquakes in New Zealand, Mw9.0earthquake in Japan, the identification ofvelocity pulse on Baker (2007) method is studied. The results show that the efficient ofvelocity pulse extraction is subject to the limitation of the wavelet base, the condition ofPGV more than30cm/s looks unreasonable. The Empirical Mode Decomposition and thelow-pass filter are presented to identify more than one velocity pulse, which get rid ofthe assumption of the pulse shape subjectively.