*4.3. Post-Processing for Virtual Shot Gathers*

Although preprocessing the raw field data before cross-correlation could effectively enhance the surface waves, the surface waves on the virtual shot gather were still concealed by the strong transverse axis energy. Determining how to eliminate the strong transverse axis energy, improve the S/N of the virtual shot gather, and further enhance the surface waves is of great significance for subsequent inversion and imaging.

Baig et al. [34] proposed several methods to improve the fidelity of noise crosscorrelation based on the discrete orthogonal S transform, which showed that time–frequency denoising of correlograms (i.e., virtual shot gathers) can alleviate this problem. The noise or signal removal algorithm based on SS-CWT proposed by Mousavi et al. [25] can also be applied to the virtual shot gather after cross-correlation to denoise correlograms and improve the S/N. Figure 7 shows the virtual shot gather (Figure 7a) and its dispersion curve image (Figure 7b) after denoising the correlogram of Figure 5a using the noise or signal removal algorithm based on SS-CWT. Comparing Figure 7a with Figure 5a, the strong transverse axis energy on the correlogram was effectively eliminated, and the surface waves were highlighted. The dispersion curve in Figure 7b (red part) is more concentrated and continuous than that in Figure 5b.

**Figure 7.** (**a**) Virtual shot gather after denoising the correlogram of Figure 5a and (**b**) the corresponding dispersion curve image.

The above research showed that the noise or signal removal algorithm based on SS-CWT could effectively enhance the surface waves after preprocessing the raw field data. At the same time, the method could also denoise the correlogram to further enhance the surface waves. Therefore, for the passive seismic datasets acquired in Sichuan, we first preprocessed the raw field data, then used the separated noise records to perform cross-correlation calculations to form the virtual shot gathers, and finally post-processed the correlograms, thus obtaining the final virtual shot gather, shown in Figure 8a, and its corresponding dispersion curve image (Figure 8b).

**Figure 8.** (**a**) Virtual shot gather after preprocessing and post-processing and (**b**) the corresponding dispersion curve image.

Comparing Figure 8 with Figures 5–7, we observed that the strong transverse axis energy in the virtual shot gather (Figure 8a) was eliminated, and the surface waves were richer and clearer; the information of the low-frequency part in the dispersion map shown in Figure 8b is prominent, and the dispersion curve is clear and continuous.
