Extending MAZ PSDM velocity model building to land context using controlled beam

migration, a case study Olivier Hermant*, Jean-Paul Gruffeille, Serge Zimine, Sylvain Navion, CGGVeritas

Summary

3D marine and land dense wide-azimuth (WAZ) and multi-

azimuth (MAZ) acquisition and processing has taken off in

the industry with the associated development of specific

tools for WAZ/MAZ processing. For land data, wide-

azimuth information actually exists since the early 3D

acquisition as land data is wide-azimuth by nature.

However and unless we talk about recent dense land

acquisition with short distance between source and receiver

lines and very high density of receivers and source points,

land data is usually limited by its trace density and its noise

content. By gathering different techniques such as

Common Offset Vector (COV), Controlled Beam

Migration (CBM) and MAZ tomography, we present in

this paper a workflow allowing the use of recent WAZ

processing tools in land context without requirements of

high trace density and the benefits of using such sequence

on the data quality.

Introduction

CBM has proven to be one of the tools of excellence for

Pre-Stack Depth Migration velocity model building, filling

in the gap between Kirchhoff migration – of flexible use

but single arrival only – and wave-equation based

algorithms (WEM/RTM) – fully handling multi-arrivals

but still time consuming (Gray et al., 2003, Vetle et al.,

2008) – and also reducing the noise content and providing

better imaging.

Beam migrations (and CBM) require the input data to have

consistent offset and azimuth information. In land (or

OBC) context, this implies using vector binning in the

processing sequence prior to migration. By running CBM

in COV domain, we preserve the azimuth information

during the migration and have full benefit of multi-pathing

inherent to beam migrations together with noise reduction

effect of CBM. We thus have the ability to pick the

residual move-out (RMO) on the common image gathers

(CIG) as a function of offset and azimuth – either using

azimuth sectors (discrete number of azimuth, equivalent to

a MAZ context) or in a WAZ sense (full 3D offset and

azimuth RMO picking).

These MAZ or WAZ RMO picks are then used as input for

non-linear slope tomography, tomography engine used for

updating the PSDM velocity model.

In marine environment, MAZ seismic acquisition and

MAZ tomography are extensively used in Nile Delta with

considerable improvements compared to narrow azimuth

data (NAZ) (Keggin et al., Rietveld et al., 2007, Gruffeille

et al., 2009). Based on a case study using data from

onshore Nile Delta and from a velocity model update point

of view, we will detail in the following article how the use

of COV and CBM can allow the use of MAZ tomography

in a similar approach as it is used in marine environment

and with the same potential of uplift on the final image. On

the imaging side, we will also show the benefits of using

COV CBM compared to standard Kirchhoff depth

migration.

Offset Vector Tiling for CBM

As mentioned earlier, CBM requires input data with

consistent offset and azimuth information. The dataset

used in this example is land data, with orthogonal shooting

acquisition layout and relatively sparse source and receiver

line spacing (350m between receiver lines and 350m

between source lines, nominal fold 66). To honour CBM

requirements, this dataset was split into COVs, each of

them having homogeneous offset and azimuth distribution

(Zimine et al., 2010), and minimum holes or fold

variations due to acquisition irregularities (Vermeer,

2002). Later on, CBM is performed independently for each

COV volume so that after migration, all traces belonging

to a CIG still have the offset and azimuth information

available.

In addition to this, after vector binning, further processing

can be applied such as data regularisation, bin centering

and trace interpolation, preserving the azimuth and offset

information.

To illustrate results obtained with this processing

sequence, Figure 1 is an overlay of seismic and velocity

Figure 1: Depth seismic section overlayed with migration

velocity model.

δ≈5%, ε≈7%

δ≈14%, ε≈19%

δ=0%, ε=0%

MAZ model building in land context using CBM

model. This is a TTI PSDM velocity model, for practical

reasons only the principal velocity (velocity along the tilt

axis) is displayed. The tilt axis was extracted from the

seismic (structural dips) and the anisotropic parameters

used are Thomsen ε and δ.

Due to large lateral velocity variations and strong

anisotropy between 3 km and 4 km deep and a velocity

inversion below that, some multi-pathing is expected in the

deeper part of the section.

Figure 2 shows that incremental uplift can be obtained

going from Kirchhoff PSDM to COV CBM and COV with

data regularization and CBM (panels from left to right,

respectively).

MAZ move out analysis on migrated gathers

Let us focus now on the analysis of CBM migrated gathers.

As each COV has been migrated separately, we can sort

the gathers using the concept of ‘snail’ gather (Lecerf et

al., 2009), i.e. by increasing offset ranges and within each

offset range by increasing azimuth (figure 3).

As described by Zimine et al., 2010, the wobbling effect

highlights a variation of the move-out with the azimuth. In

this example the cycles of the wobbling effect are not

always complete: although data is WAZ by nature, it is not

dense WAZ and the azimuth and offset distribution is not

as rich as it would be on dense WAZ acquisitions. One of

the goals of this study is to assess whether such data can be

used for MAZ tomography or not. Is the trace density

limiting the quality of the RMO picking needed for

velocity model update?

After observing the wobbling effect on the migrated

gathers, we looked at dividing the COV CBM gathers in

azimuth sectors. The aim is to pick RMO independently on

each azimuth sector.

Figure 4 shows some migrated gathers separated in 3

azimuth sectors, both CBM and Kirchhoff gathers are

displayed. The comparison between CBM and Kirchhoff

gathers shows the benefit of CBM in terms of signal to

noise ratio. Although the offsets are not evenly distributed

amongst the 3 sectors, the number of traces in each sector

is sufficient to perform automatic RMO picking on CBM

gathers but would be difficult on Kirchhoff gathers due to

Figure 3: COV CBM gather sorted in snail gather (Lecerf et al.,

2009). The wobbling effect observed on these gathers indicates

azimuthal variation of the move-out.

Figure 2: Kirchhoff PSDM (left), COV CBM (middle) and COV CBM using data regularisation in COV domain (right). All displays are in depth

with AGC applied. Note the improvement in signal to noise ratio and continuity of deeper events when comparing Kirchhoff and CBM stacks,

with additional imaging improvements when applying regularisation of COV volumes prior to CBM.

MAZ model building in land context using CBM

the level of noise and lack of continuity of the seismic

events.

Similar to the wobbling effect shown on the snail gathers

(Figure 3), we can observe on these gathers some

variations of the move-out from an azimuth sector to the

next, which stresses the importance of taking the azimuth

into account during the tomographic update.

On Figure 5 are displayed the stacks of the different

azimuth sectors (CBM stack of gathers from Figure 4). The

stacks show the illumination variations in the sub-surface

with the azimuth, described as wave path redundancy and

complementary information between the different azimuths

(Montel et al., 2010) beneficiating to MAZ tomography.

MAZ RMO picking for non-linear tomography

We update the velocity model using non-linear slope

tomography (Guillaume et al., 2001, 2008) and its

extension to MAZ datasets (Montel et al., 2010). The use

of locally coherent events is adapted to dense volumetric

picking. In MAZ context, the dense automatic picking can

be done independently for each azimuth sector, using 4th

order polynomial function for the picking, well adapted in

Figure 4: PSDM gathers divided in 3 azimuth sectors, CBM on the left, Kirchhoff on the right. The azimuth are not evenly distributed with the offset, the 2nd sector having a limited range of offset. Note the variations of the move-out from a sector to the next and the improved signal to

noise ratio on CBM gathers compared to Kirchhoff gathers

Figure 5: CBM stacks of each azimuth sector. Sector 0-60 degrees (left), 61-120 degrees (middle) and 121-180 degrees (right). Displays are in

depth, with AGC applied. Note how the energy is different from an azimuth sector to the next, each of them illuminating different areas in the

sub-surface. This effect mainly occurs below the high amplitude event at 3km, where the geology becomes more structurally complex.

MAZ model building in land context using CBM

case of presence of anisotropy as it handles complex RMO

curves (Figure 6).

In addition to the RMO picks, the other attributes needed

for the non-linear tomography are extracted from the

seismic stacks: dips and skeleton are computed for each

azimuth sector. In the non-linear tomography process

(Figure 7), the kinematic de-migration is performed for

each azimuth sector and the resulting invariants

(Guillaume et al., 2001) are merged together prior to

tomographic update, ensuring the preservation of the

azimuth information.

Quality control of the RMO picking is shown on Figure 8,

with the overlay of CBM gathers and the picked RMO

represented by a 4th order polynomial function.

Conclusions

The combination of CBM with offset vector tiling process

on land data with sparse acquisition layout demonstrates

the benefit of CBM for reducing the amount of migration

noise compared to Kirchhoff pre-stack depth migration. As

a result we can afford separating the data in azimuth

sectors preserving a reasonable signal to noise ratio in each

of them and thus allowing accurate RMO picking for

velocity model update through non-linear slope

tomography in a MAZ context.

This opens further possibilities for re-processing non-dense

land data with full advantage of taking into account the

WAZ nature of such datasets.

Acknowledgements

We thank RWE Dea Egypt for granting us permission to

show these field data examples and publishing results.

We also thank CGGVeritas for the authorization to present

this work and our colleagues from R&D and Processing for

fruitful discussions during the life of this project.

Figure 7: Non-linear slope tomography

Figure 6: (C2,C4) linear combination used for RMO polynomial

picking (h being the offset)

Figure 8: Picked RMO curves overlaid with the CBM gathers for

each azimuth sector.

References

Gray, S., C. Notfors, N. Bleistein, 2003. Imaging using multi-arrivals: Gaussian beams or multi-arrival Kirchhoff?. CSEG/CSPG

Joint Conference 2003: “Partners in a New Environment”, Calgary

Gruffeille, J.P., A. Ramadan, H. Darmen, S. Fairhead, D. Maguire, A. Speeding, A. El Fattah, 2009. Exploring Oligocene targets

using Multi-Azimuth Acquisition and latest processing technologies in MAZ context, EAGE, Workshop proceedings, “Subsalt

Imaging Workshop: Focus on Azimuth”, (15-18 November 2009, Cairo), SS32.

Guillaume, P., Audebert, F., Berthet, P., David, B., Herrenschmidt, A. and Zhang, X., 2001, 3-D finite-offset tomographic

inversion of CRP-scan data, with or without anisotropy, 71st Ann. Internat. Mtg.: Soc. of Expl. Geophys., 718 – 721.

Guillaume, P., G. Lambaré, O. Leblanc, P. Mitouard, J. Le Moigne, J.-P. Montel, T. Prescott, R. Siliqi, N. Vidal, X. Zhang and S.

Zimine Kinematic invariants: an efficient and flexible approach for velocity model building, 78th annual SEG meeting, Las

Vegas 2008, SEG workshop “Advanced velocity model building techniques for depth imaging”.

Keggin, J. and al, 2007, Multi-azimuth 3D provides robust improvements in Nile Delta seismic imaging: First Break, 25(3), 47-

54.

Lecerf, D., S. Navion, J.L. Boelle, A. Belmokhtar, A. Ladmek, 2009. Azimuthal Residual Velocity Analysis in Offset Vector for

WAZ Imaging, EAGE extended abstracts, V013.

Montel J.P. P. Guillaume, G. Lambaré and O. Leblanc, 2010. Non-linear slope tomography: extension to MAZ and WAZ

configurations, 72nd EAGE Barcelona.

Rietveld, W. and al, 2007, Multi-azimuth towed streamer 3D seismic in the Nile Delta, Egypt: processing solutions: First Break,

25(3), 55-62.

Vermeer, G., J., O., 2002, Processing with offset-vector-tile gathers: 63rd EAGE Amsterdam.

Vinje, V., Roberts, G. and Taylor, R., 2008, Controlled beam migration: a versatile structural imaging tool. First Break, 26(7),

109-113.

Zimine, S., G. Lambaré, P. Guillaume, J.P. Montel, J.P. Touré, N. Deladerrière, X. Zhang, A. Prescott, D. Lecerf, S. Navion, J.L.

Boelle, A. Belmokhtar and A. Ladmek, 2010. Can we solve RMO azimuthal variations for land WAZ, with a non-linear

tomography solution in depth? A case history. 72nd EAGE Barcelona.