99 4. Near-surface reflection surveying Shot points recorded for near-surface seismic surveys are usually corrupted by surface waves such as pseudo-Rayleigh waves. For seismic imaging based on reflected waves, it is necessary to be able to separate weak reflected events from high energy surface waves. Wave separation is a crucial step in the processing sequence. We describe here the benefit of combining two different wave separation methods to remove the energetic wavefield. The conventional F-K method is used to filter surface waves and converted refracted waves. The SVD method (Singular Value Decomposition) is then used to extract refracted waves. The different steps of the processing sequence are: amplitude recovery, deconvolution by spectrum equalization, wave separation by SVD and F-K filters, normal moveout (NMO) with constant velocity for quality control. The shot point presented here is an end-on spread shot composed of 96 traces. The distance between 2 adjacent geophones was 1 m. The source was a weight dropper (see Figure 2.6-c in the “Refraction surveying” chapter). There was no data filtering at the acquisition, consequently the shot was highly corrupted by surface waves (pseudo Rayleigh modes). This shot type is often called a noise profile. It can be used for the analysis of surface waves and also to define the acquisition parameters for near-surface 3D acquisition. Figures 4.7 to 4.8 show the step-by-step processing sequence of the noise profile. At each stage the data are shown both in the time-distance domain, and in the frequency-wavenumber domain (f-k domain). Figure 4.7-a shows the raw shot. The seismic trace close to the source is saturated. We observed a very strong attenuation of seismic amplitudes with the offset. After amplitude compensation (Figure 4.7-b), the direct wave, refracted waves, air wave and surface waves were clearly visible. The f-k amplitude spectrum shows that most of the energy is limited in wave number up to 0.25 c/m. Consequently, a geophone interval of 2m allowed the data to be recorded without spatial aliasing. A deconvolution process was applied to the data to increase the vertical resolution by spectral balancing and to facilitate the wave separation (Figure 4-d). The wave separation process involves the extraction of a wave by an apparent velocity filter defined in the f-k domain and then the subtraction of the estimated wave from the dataset to obtain a residual section. The process is carried out iteratively for different waves or seismic events. Figure 4.7-d shows the estimation of the air wave and the Rayleigh wave. The associated residual section is shown in Figure 4.8-a. On the 2D amplitude spectrum, we note that the energy is concentrated in the 0 to 0.2 c/m wave number interval. It is also possible to see that the air wave is aliased for frequencies larger than 200 Hz and appears with negative apparent velocities. The events with negative apparent velocities are shown in Figure 4.8-b, while the associated residual section (Figure 4.8-c) mainly contains the refracted events (Figure 4.8-d). The residual section associated with the refracted events shows events of very weak amplitude with high apparent velocities in the 60 to 150 ms time interval (Figure 4.9-a). These events are reflected events. On the same section, we can observe low apparent velocity events which are residues of direct waves and air waves (Figure 4.9-b). The residual section associated with the low apparent velocity events shows reflected events (Figure 4.9-c), which are flattened after NMO corrections (Figure 4.9-d).
RkJQdWJsaXNoZXIy NjA3NzQ=