DYNAMIC SPECTRAL ANALYSIS OF INFRASONIC SIGNAL WITH HILBERT TRANSFORM

In the past we used to analyse the dynamic spectrum of infrasonic waves with FFT1, but which is not always suitable. In this paper another method is proposed to analyse infrasonic waves. This method is that, after digital filtering of the signal and Hilbert transform, the envelope of the signal in all the channels is evaluated, the bandwidth of the filters is 1/12-octave. In this paper the computation of envelope with Hilbert transform and 1/12-octave filtering are discussed. Then a synthetic infrasonic signal generated by a large nuclear explosion is analysed with two methods, and their results are compared. It is shown that, in the dynamic spectrum with Hilbert transform, not only higher resolution can be obtained, the main frequency regions of different modes can be separated, but also waveform can be preliminarily obtained for a few of lower modes. Thus it seems to that in some cases this method is more efficient in discovering the property of an infrasonic signal.