243 8. Seismic anisotropy applied to geothermal prospection For HTI media, rock physics models, such as Mori-Tanaka (Mori and Tanaka, 1973), allow computing the elastic stiffness of an inclusion model, mixing some fracture apertures/orientations. Synthetic seismic data generated in 1D by such models allows us to draw general observations: • Anisotropy is not detected at short offsets. • If one fracture set is present, with a given orientation, a large fracture density and large fracture lengths, it leads to a measurable anisotropy in the seismic at large offset (or angle, around 30°). • Two equivalent perpendicular fracture sets lead to isotropic result and kill the anisotropy effect; while changing the balance between both sets, the anisotropy intensifies at large offset or incidence angle. In practice, not only faults and fractures can be detected using HTI media approximation. Any brutal and oriented change affecting the wave, such as a lithology or porosity change, can generate similar anisotropy. In addition, the lateral resolution is also a key factor: If an element is wide enough to be detected, regardless of the azimuth, no anisotropy will be induced. 8.1.2 Azimuthal stacking and required processing In the previous chapter, the partial stacking was introduced to generate full-stack or angle-stacks from gathers. Another stacking method, to detect anisotropy, is possible with Wide Azimuth (WAZ) seismic acquisition and associated gathers using the azimuthal key, representing the direction between the line source/receptor and the north. As for the previous application, the gathers must be “amplitude preserved”. The response of the signal is expected to be symmetrical: exchanging the location between the source and the receptor should lead to a similar signal. An azimuthal range of 0-30° is then equivalent to a range of 180–210°. This technique allows more traces to be involved during stacking, and, therefore, to reduce the noise content of the azimuthal stacks. For the VVAZ approach, different alignment processes might destroy the expected anomalies, especially: • All the azimuth dependent velocity picking or RNMO. • The trim-statics, which is a process that aligns seismic events using dynamic shifts. • Filters, such as Radon or F-K filters, might not be adapted to preserve the azimuthal information. As the anisotropy is mostly detectable for the large offsets/angles (Chérel et al., 2010), these must be considered during the stack (even up to very large offsets, further than Aki-Richards classical limitations for elastic inversion workflows). This way, the chance to detect anisotropy would significantly increase.
RkJQdWJsaXNoZXIy NjA3NzQ=