Rayleigh wave processing

During the same campaign, the geophone layout was used to record artificially generated Rayleigh waves. Sources were placed with an offset of 20 m in order to avoid near-field effects on the closest receivers. Two kinds of sources were used: hammer shots like in the preceding section and explosive loads (100 gr of black powder) buried at about 0.8 m deep. Eight shots were recorded, corresponding to the two profiles, the two source types, and the two extremities of profiles. The averages and the standard deviations of the frequency spectra observed for the two sources are compared to the ambient noise level in figure 6.24. The amplitude for the explosive shots is about 25 times greater than the amplitude for the hammer shots. The energy level of the ambient vibrations is so high that the results from hammer shots might be valid only inside a narrow frequency band between 15 and 25 Hz. On the other hand, explosive shots are far above the ambient noise for all frequencies between 6 and 50 Hz.

Figure 6.24: Triggered surface waves along $P-S_V$ profiles North-South and East-West. Average spectra (plain lines) and standard deviations (dashed lines) for 24 receivers recording a black powder shot situated at 20 m from the first sensor (thick black lines) and a hammer shot (grey lines) and the ambient vibrations recorded with the same sensors (thin black lines).

r0cm

Figure 6.25: Triggered surface waves along $P-S_V$ profiles North-South and East-West. (a) to (d) Frequency-wavenumber semblance maps for the different shots: (a) East explosive shot and (b) East hammer shot on EW profile, (c) South explosive shot and (d) South hammer shot on NS profile. (e) Picking of the maximum apparent velocity for all source locations and types (8 curves, thin lines for hammers and thick lines for explosives).

For each shot and for all frequency bands, a frequency-wavenumber semblance is calculated for the linear arrays of sensors. The technique is exactly the same as for the processing of microtremor arrays (section 1.2.3), except that only one time window is processed. The apparent velocity is deduced for each frequency band. The calculated semblance plots are shown for two shot positions in figures 6.25(a) to 6.25(d). Figures (a) and (c) are for explosive shots and figures (b) and (d) for hammer shots. The consistency of the measured dispersion curve (maximum of semblance) checked for all eight sources in figure 6.25(e) is remarkable. All plots are cut between 8 and 40 Hz which is inside the valid interval for explosive shots but, amazingly, outside the hammer shot validity range. One reason could be that the ambient noise is predominantly made of surface waves, leading to a global coherency of the semblance function. Below 10 Hz, for both explosive and hammer shots, the uncertainties over the velocity estimates drastically increase (figures 6.25(a) to 6.25(d)).

The Rayleigh dispersion curve is not directly inverted here to obtain the $V_s$ or $V_p$ profile. It is used in the next sections, comparing to array results.