GNGTS 2015 - Atti del 34° Convegno Nazionale

128 GNGTS 2015 S essione 3.3 Timemigration vs. normal moveout. Timemigration proves to be a great tool for processing data coming from areas where lateral velocity changes are not too severe, but structures are complex. Time migration has the effect of moving dipping events on a seismic surface from apparent locations to their true locations in time. The resulting image is shown in terms of travel-time rather than depth, and must then be converted to depth with an accurate velocity model to be compared to well logs. When velocity variations are sufficiently strong to create non-hyperbolic diffraction surfaces, canonical time migration cannot focus the data properly, and depth migration could become necessary. However, if interval velocity estimates are inaccurate, depth migration not only mispositions the reflectors in depth but also misfocuses the data. Its robustness makes time migration an essential imaging tool. However, it has to be applied within the limitation imposed by its underlying assumptions. Time migration in his isotropic version assumes that the diffraction surfaces are hyperbolic, and that the corresponding time-shift can be reasonably well described by the following equation: (1) where s, r and m represent source, receiver and image point coordinates on the surface respectively. Time migration is very sensitive to the choice of the velocity function and the irregular sampling of the data. An approximate method for imaging prestack data that is less expensive and often more robust than full prestack migration is based on the transformation of prestack data to equivalent zero-offset data. This is accomplished by the application of normal moveout (NMO): (2) At first order, NMO transforms data collected at a finite offset to equivalent zero-offset data. The transformation to zero-offset is only kinematic and approximate; even in constant velocity, the NMO time is exactly equivalent to zero-offset time only when reflectors have no dip component along the trace source-receiver azimuth. The NMO and stack sequence can be seen as an approximate partial prestack time migration that focuses data along the offset axes. The only velocity information that NMO and stack needs to focus the data along offsets is the average velocity corresponding to the midpoint which can be directly measured from the data. Because NMO is a trace-to-trace transformation, with no spreading of energy across midpoints, NMO and stack is immune from operator aliasing across midpoints (Biondi, 2006). Anisotropy. Anisotropic models are recognized as a more realistic representation of the subsurface where complex geological environment exists. However, anisotropic model building is still a challenging problem in the industry. Currently, many seismic processing and inversion methods utilize anisotropic models, thus providing a significant enhancement over the seismic imaging quality and resolution. The existing anisotropic model-building schemes are mostly based on measuring the non- hyperbolic moveout along the traveltime curve to flatten the common image gathers (Li and Biondi, 2011). Considering horizontal layers only, the medium can be defined as vertical transverse isotropic (VTI). In case of weak anisotropy the NMO curve can be approximated by Taylor expansion using an extra term depending on the anisotropy parameter η (Alkhalifah and Tsvankin, 1995). Time migration using such approximation can be rewritten as: