Correction of highly noisy strong motion records using a modified wavelet de-noising method

During the past earthquakes, many valuable acceleration time histories were recorded by analog and digital accelerometers. These records are important sources of information in the field of earthquake engineering and engineering seismology. However, a large number of these records are contaminated by noise and it is necessary to correct them for practical applications. On the other hand, only a few records can be corrected using conventional filtering because of mathematical limitations of the method. However, advances in the field of time–frequency analysis and wavelet transform theory provide useful non-linear and adaptive de-noising methods for removing of non-stationary and high-energy noise from the recorded signals. In this paper, the characteristics and capabilities of the modified non-linear adaptive wavelet de-noising method are examined for correction of highly noisy strong motion records. In the frequency domain, it is shown that this method can attenuate the noise in the whole frequency range of engineering interest while in the time domain it can detect and remove non-stationary noise. In addition, the displacement response spectra of these wavelet de-noised records are more stable than conventional filtered records with respect to different correction functions. It is found that a large number of noisy acceleration records that are usually discarded from sets of records used for estimating the ground motions can be corrected using this new method.