CRS technique for advanced prestack merging and regularisation of vintage 3D seismic dataGuido Gierse* ,Dennis Otto, Arnim Berhorst, Henning Trappe, Juergen Pruessmann, TEEC

Summary

The merging of seismic data of different origin is a

common task in the reprocessing of vintage 3D seismic

data. In contrast to poststack merging, the prestack merging

is rewarded by the much broader possibilities of prestack

migration and analysis techniques but requires a larger

effort to adjust different acquisition and bin geometries,

and to interpolate missing data in the binning grid of the

merged dataset. In this case study, a new strategy is

proposed using Common-Reflection-Surface (CRS) partial

stacking for both, the merging and the regularisation of the

prestack data from two 3D marine surveys. The acquisition

already provided some irregularities in the CMP and offset

coverage of both surveys, which are increased by the

adjustment of the binning grids requiring a smaller grid cell

in one of the datasets. In a addition, the overlap zone of the

two surveys exhibits a general decrease of the coverage.

The data which are missing in the regular CMP/offset grid

of the merged dataset are recovered by partial CRS

stacking of original traces in the CMP/offset vicinity of a

missing regular trace. This data mapping benefits from the

detailed event description in the CRS attributes derived in

the CRS zero-offset stacking workflow. It combines a dip-

consistent interpolation of the prestack data with a

significant increase of the signal-to-noise ratio as part of

the partial CRS stacking.

Introduction

Modern acquisition equipment, and increasing processing

capacities based on high-performance IT technology have

stimulated a steady growth of project sizes in 3D seismic

surveying. This trend to larger units has also influenced the

reprocessing of old 3D seismic data, that often had been

acquired in much smaller patches. The merging of several

small or medium 3D seismic surveys of different vintages

has thus become a common task in seismic exploration

projects. Since contemporary processing sequences have

replaced former poststack imaging by prestack migration

techniques the merge is generally performed in prestack

domain.

Before merging, different acquisition footprints, signal

characteristics, amplitude levels, and static shifts are

commonly adjusted separately in the individual 3D surveys.

The prestack merging then aims at a maximum

homogeneity of the resulting dataset, not only comprising

the similarity of the seismic events in traces from different

sources, but also the structure of the dataset. With a

consistent regularisation throughout the dataset prestack

migration is expected to minimize migration noise, and

produce the best results. This case study concentrates on

the aspect of adapting the dataset structure in a 3D seismic

merge project, and proposes a new workflow based on CRS

regularisation.

CRS interpolation strategy

The CRS method, or Common-Reflection-Surface method,

was originally developed by Hubral et al. (1999), Mann et

al. (1999), and Jaeger et al. (2001) within the concept of

macro-model independent imaging (e.g. Gelchinsky 1988).

CRS zero-offset stacking assumes local reflector elements

with dip and curvature in the subsurface that give rise to the

seismic reflections. The corresponding CRS stacking

parameters, the so-called CRS-attributes, accordingly

comprise the wavefield dip together with wavefront

curvatures observed at the surface. They define hyperbolic

CRS stacking surfaces that extend across several CMP

locations, and thus collect high-fold contributions from the

prestack data.

The CRS attributes are optimized locally for each point of

the image, thus providing a detailed kinematic description

of the seismic events in the data that can also be used for

mapping seismic data to a regular grid of traces. Event data

from original traces in the vicinity of a regular trace is

mapped to that regular trace by dip-consistent partial CRS

stacking, based on the CRS attributes.

This CRS interpolation strategy has proven to be a suitable

tool for regularizing CMP and offset coverage within single

3D seismic datasets (Gierse et al. 2009). Similar

regularization techniques based on the local measurement

of time dips and curvatures have been successfully

performed in various data domains by Hoecht et al., 2009,

but without associated model assumptions of local curved

and dipping reflector elements from CRS zero-offset

imaging.

The merging and regularisation of vintage 3D seismic data

with various acquisition designs generally includes more

complex interpolation tasks than single 3D seismic

datasets. Incompatible binning grids due to different

acquisition parameters lead to large portions of empty grid

cells after regridding.

Merging and regularisation of two 3D seismic surveys

The 3D seismic data to be merged in this case study

comprised two marine surveys from the Norwegian North

Sea which differed not only in subsurface fold but also in

acquisition direction and bin cell size. The fold maps and

CRS technique for advanced prestack merging and regularisation of vintage 3D seismic data

Figure 1 - Original CMP fold maps of two marine surveys to be merged. Both maps cover the same area at the same scale showing the strong

fold variations of the surveys. The corresponding bin cell geometries are added as dotted rectanglar frames at strongly exagerated scales. A merge

example line illustrates the direction of data extraction before and after merging in the gather and stack displays of Figurey 2 and 3, respectively.

the original bin cells are displayed in Figure 1 for these

datasets termed Survey 1, and Survey 2, respectively. The

inline directions of the two surveys are perpendicular to

each other, as well as the original rectangular bin cells. The

inline width of 12.5 m of these cells had been defined

previously after trace decimation from the acquisition width

of 6.25 m. In addition to the different orientations of the

inlines, the surveys exhibit strongly inhomogeneous fold

distributions in the fold maps of Figure 1. Strong feathering

and irregular acquisition required a data regularisation even

for Survey 1 where the bin cell was retained in the merge.

Survey 1 incorporated a bin cell of 12.5 m X 25.0 m which

was adopted for data interpolation and regularisation in the

merge project. This merge geometry strongly contrasted

with Survey 2 showing a bin cell of 37.5 m X 12.5 m due to

the different orientation and separation of the streamers.

The regridding in the merge project intended to refine the

grid interval in the first dimension from 37.5 m by a factor

1/3 to 12.5 m, and to coarsen the interval in the second

dimension from 12.5 m by a factor 2 to 25.0 m.

The prestack merge and interpolation procedure is

illustrated along a line following the inline direction of

Survey 1 like the example line in Figure 1. Simple re-

binning of Survey 2 to the bin grid of Survey 1 produced an

irregular and sparse data distribution in the CMP gathers of

the new inline and crossline grid, with a large proportion of

empty grid cells, and of grid cells with large offset gaps.

Figure 2a (top) shows some CMP gathers after rebinning in

the overlap zone, and in the adjacent regions of both

surveys. Survey 1 which enters the merge procedure

unchanged provides a reasonable offset distribution but also

exhibits some missing offsets due to the irregular

acquisition fold as shown in Figure 1. In the overlap zone,

only partial offset ranges are covered within the CMP

gathers, or there is hardly any data at all in these gathers. In

Survey 2 the rebinning fills each CMP gathers at several

small offset ranges only in which the data happens to fall

into the redefined smaller grid cells.

CRS technique for advanced prestack merging and regularisation of vintage 3D seismic data

Figure 2b (bottom)

displays the correspon-

ding CRS gathers in

which missing data

were reconstructed by

partial CRS stacking of

original data from the

vicinity of each desired

new CMP / offset loca-

tion. The partial CRS

stacking completely

filled the large data

gaps in the low-fold

overlap zone and in the

regridded Survey 2,

and also compensated

for the irregular offset

distribution of Survey

1. In addition, the

partial stacking strong-

ly increased the signal-

to-noise level.

In this CRS data

interpolation strategy,

the missing data were

reconstructed from the

kinematic event infor-

mation supplied by the

CRS attributes. Unlike

conventional grid map-

ping and regularisation

methods, this CRS

technique did not

require any interpolat-

ion between neighbor-

ing shots, or flexible

binning techniques that

degrade dip and

resolution.

NMO stacks of the

prestack data before

and after this regulari-

sation served as a

quality control as

shown in Figure 3. The

near-surface data gaps,

and some fully missing

traces were consistent-

ly filled by the CRS

interpolation strategy.

As expected, the CRS

technique did not

Figure 2a – CMP gathers after data regridding to the merge grid. Note the irregular offset coverage.

Figure 2b - CRS gathers corresponding to CMP gathers of Figure 2a after partial CRS stacking.

Survey 1 Survey 2Overlap zone

Survey 1 Survey 2Overlap zone

CRS technique for advanced prestack merging and regularisation of vintage 3D seismic data

imply any loss of resolution. On

the contrary, the image was even

improved due to the compensation

for some irregularities and fold

variation in the original data, and

due to the noise suppression

during the partial CRS stacking.

Conclusions

This case study demonstrates the

successful prestack merging of

two 3D marine datasets using a

CRS interpolation and

regularisation strategy. The

strategy is based on the CRS

attributes from the CRS zero-

offset stacking workflow that

represent a detailed database of the

local kinematic behaviour of

seismic events in multicoverage

data. These attributes allow a local

mapping of existing seismic data

to a regular merge grid in the

CMP/offset domain. The strength

of this CRS interpolation and

regularisation technique is

emphasized in areas where the

merge grid implies a refinement of

the original bin grid, and an

associated deterioration of the

original CMP/offset coverage.

Partial CRS stacking of original

data in the vicinity of each regular

CMP/offset location results in so-

called CRS gathers. These CRS

prestack gathers show a uniform

CMP/offset coverage with a

complete recovery of missing

prestack data, and a significant

increase of the signal-to-noise

ratio. This merged and regularized

prestack data is considered as an

ideal input for prestack migration

which is presently performed.

Acknowledgements

We thank Wintershall Norge ASA

for their kind permission to

present their data.

Figure 3a – Zero-offset stack of CMP gathers as in Figure 2a after data regridding to merge grid.

Figure 3b – Zero-offset stack of CRS gathers as in Figure 2b after partial CRS stacking.

Survey 1 Survey 2Overlap zone

Survey 1 Survey 2Overlap zone