GNGTS 2014 - Atti del 33° Convegno Nazionale

GNGTS 2014 S essione 3.1 61 Water layer multiple migration does not provide wider illumination, however the increased fold of coverage promotes better signal-to-noise ratio in the same region where the conventional VSP imaging works. Moreover, the reflection angles of recorded multiples are mostly smaller than those of the corresponding primaries: smaller reflection angles provide higher vertical resolution and can sometimes be less prone to defocusing. The complete flowchart describing the whole procedure is shown in figure 1a: the ability of imaging water layer multiples through a simple source geometry transform compares favorably to approaches explicitly or implicitly based on interferometric or redatuming techniques (Jiang et al. , 2007), although the locally planar sea-bottom hypothesis may be not completely fulfilled. As first order water layer multiples add a round-trip to corresponding primary reflection ray-paths, a lower signal-to-noise ratio is expected because of greater geometrical spreading. However, other attenuation losses can be safely ignored because the additional ray-path is confined in the water-layer only. Furthermore, the effects on recorded amplitudes of the reflection at sea-bottom has been ignored so far, which implies that a constant sea-bottom reflection coefficient is implicitly assumed. Thus, imaging results should be considered kinematically correct but one should not trust on retrieved amplitudes. It must be also noted that in principle the proposed approach can be applied to higher-order water-layer multiple reflections by iterating the procedure. Synthetic data example. The proposed procedure has been tested with a simulated 2D Walkaway VSP dataset. The velocity model has been extracted from the BP2007 benchmark model (software.seg.org ): the model extends 20 km laterally and 11.25 km vertically, with a spatial sampling of Δx =6.25 m (corresponding acoustic impedance is shown as colour- coded background in Fig. 3). Seismic data have been computed by a finite difference isotropic acoustic simulation (using both velocity and density models), with staggered grids. Output time sampling is Δt =1 ms while trace duration is T =16 seconds. The dataset geometry consists of a single geophone, placed at z r =4000 m depth in the well, that is illuminated by 1601 sources at sea-surface with inter-source distance Δs =12.5 m each other (acquisition geometry is partially sketched in Fig. 3b). Density model is selected to simulate a hard water-bottom interface (mean water layer depth is w d =450 m). As the subsurface illumination is both laterally and vertically extended by the previously described receiver mirror imaging technique, results comparison is shown for “receiver mirror- imaged” results only, although the analysis and comparison of “conventional” (non mirrored- receiver) results lead to the same conclusions. Pre-stack depth migration is performed by a ray- based Kirchhoff migration, using a smooth velocity model. Receiver mirror-migrated result of WVSP data after Water Layer Multiple subtraction (shown in Fig. 3b) proves how imaging step succeeds to correctly recover most of the subsurface interfaces. This result, compared to the mirror-migrated image of WVSP raw data (i.e., before any interfering wavefield separation), shown in figure 3a, proves how migration stacking power cannot properly attenuate water layer multiple interference. Finally, in the proposed source double-mirror imaging result is shown in Fig. 3c. The comparison of Fig. 3c with Fig. 3b (water layer multiple free benchmark) proves that first order water layer multiple data can be coherently migrated and carries full subsurface illumination. Multiple migration residuals, highlighted by arrows and in the circled area on the migrated panels, are clearly visible on raw data migration result (Fig. 3a), while they are highly attenuated in both results shown in Figs. 3b and 3c. However, a globally lower SNR of multiple migration result suggests that multiple migration can complement, and not replace primary migration. Conclusions. In this work we have presented a simple and straightforward procedure to estimate, separate and finally image a subset of multiple reflections (water layer multiples) for WVSP surveys. Results on simulated data prove how migration of such surface-related multiple subset can be easily performed by a simple transformation of source point geometry.