An efficient detection method of the peak frequency of microtremor H/V spectral ratio by using a low-pass filter

Abstract

The present study aims at proposing a method to evaluate the peak frequency in the microtremor H/V spectra efficiently. The proposed method deals with the spectra as the time series data and applies fast Fourier transform (FFT) to the H/V spectra as the first process. In the next process, both the constant term indicating the average of the time series data and the (m+1)th harmonics and higher are removed and treated as zero. As the results, from the first to the (m)th harmonics in the spectra were remained. The remained spectra were assumed as the smoothed spectra and by applying the inverse fast Fourier transform (IFFT), the smoothed spectra without insignificant ripples were obtained. The major advantage of the proposed method is the fact that the coherent predominant peak frequency can be evaluated irrespective of the value of the peak frequency. Authors apply both the proposed method and the ordinal method to 10 observed microtremor H/V spectral ratios with different characteristics in the spectra. Predominant peak frequencies may be evaluated by the cautious evaluation of the skilled engineers without the help of the proposed method, however, it should be noticed that the predominant peak frequencies can be evaluated very easily by the proposed method.