Académique Documents
Professionnel Documents
Culture Documents
x = xm + 1 V 2t 4 m m t ( x, y ) = t m 1 +
t m x m
y = ym + 1 V 2t m 4 m m y
1 2 V 4 m
(tm )
The un-migrated reflection is mapped into depth with a normal ray map migration. This mapping requires of course a velocity-depth model.
Figure 1: Time-to-depth conversion of a migrated horizon (red) via demigration (blue) and normal ray depth migration (black).
In order to obtain an accurate velocity model for time-todepth conversion, we follow the approach of Sexton, 1998.
SEG Int'l Exposition and 74th Annual Meeting * Denver, Colorado * 10-15 October 2004
)(
T
after non-hyperbolic FullPSTM has therefore to be based on this formula, in particular for steep dips. In contrast to the hyperbolic case, this formula cannot be directly inverted to obtain the demigrated data. We implemented an iterative algorithm and used the hyperbolic solution as initial guess.
Time to depth conversion after non-hyperbolic FullPSTM
Non-hyperbolic FullPSTM
We created a simple 1-D velocity model (Figure 2) with vertical gradient and calculated analytically the migrated horizon times, V m and S m from the model parameters. This has the disadvantage that the migration parameters do not perfectly flatten the migrated gathers for large offset-depth ratios (Figure 2) but we avoid velocity and horizon picking.
It is well known from stacking velocity analysis that VTI anisotropy and fine layering generate non-hyperbolic reflection moveout (e.g. Hake et al, 1984, Alkhalifah, 1997). The seismic industry uses currently non-hyperbolic moveout approximations with different parameterizations (e.g. deBazelaire, 1988, Castle, 1994, Alkhalifah, 1997, Siliqi et al, 2001). In this paper we use the non-hyperbolic moveout formula of Castle, 1994, to demonstrate the impact of non-hyperbolic FullPSTM on time-to-depth conversion and velocity model building. The conclusions will be valid for all types of non-hyperbolic FullPSTM. The non-hyperbolic FullPSTM operator with Castles shifted hyperbola is given by
1 T ( x, y , h ) = t m 1 S m
Z t
Figure 2: Velocity model with three dipping interfaces and migrated gather after non-hyperbolic FullPSTM.
tm + 2S m tm + 2S m
( x x h )2 + ( y m y ) 2 + m 2 S mV m
2
where (x m , y m , t m ) is the Kirchhoff summation output point in the migrated volume with half-offset h, and V m , S m are the velocity and shift parameters for migration. It is assumed that the x-direction is on the source-receiver line and the y-direction perpendicular to it. The hyperbolic DSR equation is obtained by setting S m = 1 . The post-migration stack simulates the migrated zero-offset section. The non-hyperbolic zero-offset migration operator 1 T ( x, y, h = 0) = tm 1 S m
( x x + h )2 + ( y m y )2 + m 2 S mV m
2
t
Figure 3: Horizons on migrated zero-offset section (non-hyperbolic FullPSTM).
tm (x x )2 + ( ym y )2 +4 m + 2 S S mVm m still depends on the shift parameter for all non-zero migration offsets. The demigration of interpreted horizons
2
Figure 3 confirms that the analytic calculation of the horizon times is an excellent approximation to the migrated reflections after non-hyperbolic FullPSTM. Figure 4 shows the demigrated horizons on the non-migrated zero-offset section. The demigration was performed with the nonhyperbolic zero-offset migration operator. The difference between non-hyperbolic and hyperbolic demigration for all three horizons is shown in Figure 5. Note that the difference increases with dip. Figure 6 shows a detail near the deepest horizon, combining the information in Figures 4 and 5.
SEG Int'l Exposition and 74th Annual Meeting * Denver, Colorado * 10-15 October 2004
Hyperbolic demigration
90m
t
Non-hyperbolic demigration
Figure 4: Demigrated horizons on zero-offset section.
Depth [m]
t
Horizons Non-hyperbolic Demigration Hyperbolic Demigration
Figure 7: Depth conversion of demigrated horizons with exact velocity model. The horizon demigrated with the hyperbolic formula is 90 m too shallow.
pick pick Vstk Vstk (m) 1 Vstk Vstk (m) C pick O(m) = pick S S stk S stk (m) stk S stk (m) T
500ms
500 m Figure 6: Zero-offset section with horizon in migrated position (yellow line), non-hyperbolic demigration (red dots) and hyperbolic demigration (green dots).
Figure 7 shows the depth converted horizons after hyperbolic and non-hyperbolic demigration. The exact velocity model was used. The positioning error on the time section in Figure 6 translates into an error of 90 meters in depth.
Velocity model building after non-hyperbolic FullPSTM
We applied the tomographic inversion to the analytically calculated migration velocities and shift parameters, first without prior residual moveout correction or offset muting, using (a) the non-hyperbolic and (b) the hyperbolic FullPSTM kinematics. We fixed the vertical velocity gradient to the exact value (k=0.5) and obtained in case (a) V0=1516m/s and in case (b) V0=1497m/s. This result is biased by the non-corrected residual moveout (Figure 2). Although the model parameters are better restored with hyperbolic FullPSTM kinematics, the model obtained with non-hyperbolic FullPSTM kinematics provides an almost accurate depth-conversion of the deepest horizon (Figure 8). The depth error is again about 90 meters. In Figure 8, the yellow triangle is a velocity analysis point in the migrated position. Non-hyperbolic FullPSTM migrates the data of the red rays into the velocity ananlysis position. The green rays correspond to hyperbolic FullPSTM. After residual moveout correction we obtained in case (a) the accurate V0=1500m/s and in case (b) V0=1480m/s.
Application to real data
The time-to-depth conversion process is used for the tomographic stacking velocity inversion and we have shown that it is important to take the kinematics of nonhyperbolic FullPSTM into account. Another aspect of nonhyperbolic FullPSTM is that residual moveout corrections should flatten non-hyperbolic events in the migrated gathers. In order to invert the non-hyperbolic stacking velocities we adapted the objective function of Williamson and Sexton, 1999, for Castles shift parameter:
Finally, we applied the tomographic stacking velocity inversion to a marine seismic data set processed with nonhyperbolic FullPSTM. The selected depth interval is between the water-bottom and the base of a thick sedimentary formation with fine layering. Well data suggest an average a priori value of the gradient k=0.45, which was held fixed during inversion of non-hyperbolic stacking velocities and depth-markers of several wells.
SEG Int'l Exposition and 74th Annual Meeting * Denver, Colorado * 10-15 October 2004
Non-hyperbolic FullPSTM
Hyperbolic FullPSTM
isotropic
anisotropic
wells
m/s
Exact Figure 8: Deep depth horizon in inverted velocity model obtained with tomography using kinematics of non-hyperbolic (red) and hyperbolic (green) FullPSTM. The exact horizon is displayed in yellow (almost coincident just above the red horizon). Figure 10: Average interval velocities after isotropic and anisotropic inversion. The black bar indicates the range of interval velocities observed in well data.
Acknowledgements
We determined (a) an isotropic velocity layer (V0(x,y),k) by inverting only the stacking velocities, (b) an anisotropic velocity layer (V0(x,y), k, , ) using all data. Figure 9 shows the depth converted horizons for both cases with some well markers (circles). The anisotropic model can resolve the mismatch between horizon and well markers while providing realistic interval velocities which are observed in well data (Figure 10).
V(x,y) K=0.45
I would like to thank Total for the permission to publish this work and Paul Sexton (Total), Nabil Tnacheri (NTConseil), Jerme Gonnon (NTConseil) and Jean-Louis Petit (Ajilon Engineering) for their contributions.
References
isotropic
anisotropic
Figure 9: Top: Depth conversion with isotropic velocity model. The horizon is too deep for three wells markers out of four. Bottom: Depth conversion with anisotropic velocity model. The horizon matches three markers. One marker is below the horizon.
Conclusions
Accurate depth conversion of horizons picked after nonhyperbolic FullPSTM needs to take into account the nonhyperbolicity of the migration operator. We presented a method to build an anisotropic velocity model for this depth conversion problem and applied it to a real data set.
Alkhalifah, T., 1997, Velocity analysis using nonhyperbolic moveout in transversely isotropic media: Geophysics, 62, 1839-1854. deBazelaire, E., 1988, Normal moveout revisited: Inhomogeneous media and curved interfaces: Geophysics, 53, 143-157. Castle, R.J., 1994, A theory of normal moveout: Geophysics, 59, 983-999. Hake, H., Helbig, K., and Mesdag, C.S., 1984, Three-term Taylor series for t2-x2 curves over layered transversely isotropic ground: Geophys. Prosp., 32, 828-850. Marcoux, M., Godfrey, R., and Notfors, C., 1987, Migration for optimum velocity evaluation and stacking (MOVES): 49th EAGE Annual Meeting. Sexton, P., 1998, 3D Velocity-depth model building using surface seismic and well data: PhD thesis, University of Durham. Siliqi, R., Bousqui, N., and Hardouin, D., 2001, Time to move to anelliptic time processing: 71th SEG Ann. Internat. Mtg., Expanded Abstracts. Thomsen, L., 1986, Weak elastic anisotropy: Geophysics, 51, 1954-1966. Williamson, P.R, and Sexton P., 1999, Anisotropic prestack velocity inversion: A North Africa case study: 69th SEG Annual Mtg, Extended Abstracts, 1592-1595. Whitcombe, D.N., 1994, Fast model building using demigration and single-step ray migration: Geophysics, 59, 1290-1304.
SEG Int'l Exposition and 74th Annual Meeting * Denver, Colorado * 10-15 October 2004