Method

Microseisms are dominant seismic noise in the frequency range from 0.1 to 5 Hz and they are generated by pressure changes on the ocean bottom. Since understanding of the coupling mechanism between the ocean and sea floor is limited, the reliable simulation of the microseism wavefield is quite difficult. Fortunately, Dolenc and Dreger (2005) found that FDP is independent of the level of microseism excitations. This observation indicates that the H/V ratio is not dependent on the detailed source process of the microseisms because taking the H/V ratio inherently cancels the source effects and enhances the propagation effects. We have developed a simple method to simulate FDP due to microseisms or any continuous sources for given 3D velocity models. Our new method is tested by comparing the synthetic FDPs to observations from SCVSE (Figure 2.16). The method is very straightforward and consists of 5 steps. First, we compute Green's functions using a 3D finite-difference code (E3D; Larsen and Schulz, 1995). Here we assume a shallow (500m depth) vertical CLVD source as the source of the microseisms in the continental margin, and a simple Gaussian source time function is applied. Second, the Gaussian source time function is deconvolved from the synthetics. Third, in order to reduce grid-dispersion effects (e.g. Levander, 1988) a low pass filter with a corner frequency of 0.8 Hz is applied to synthetics. Fourth, continuous monochromatic sine waves for discrete frequencies over the range from 0.025 to 0.8 Hz, with an interval of 0.025 Hz, are convolved with the low-pass filtered synthetic waveforms for each station. Finally, maximum amplitudes of convolved waveforms of three components for each frequency are taken in the time window after amplitudes become stable. The final horizontal maximum is determined by taking the geometric mean of two maximum horizontal amplitudes. In this study, the definition of the H/V spectral ratio is just the ratio of maximum vertical and horizontal amplitudes as a function of frequency. Since synthetic vertical and horizontal amplitudes are not stably varying with frequency, and sometimes abnormally small vertical amplitudes cause unrealistic peaks in the H/V ratio, the moving average over 5 adjacent data points is applied to the horizontal and vertical amplitudes before taking the H/V ratio.

Figure 2.16: (a)The range of 3D models used for synthetic Green's function computation. Small triangles indicate the stations deployed during Santa Clara Valley Seismic Experiment (SCVSE). Solid circle represents assumed location of the microseism source. (b) The background shading indicates the depth of the Santa Clara Valley obtained from the 3D USGS velocity model