Common-Reflection-Surface Stack

The Common Reflection Surface (CRS) method is a macro-model independent imaging method in time domain, that was developed at the University of Karlsruhe, Germany. Unlike the conventional NMO/DMO stach, which requires a stacking velocity field as a macro-model, the CRS processing automatically determines the stacking parameters from semblance measurements in the prestack data.

The CRS stack corresponds to a subsurface model, where reflector elements - the common reflection surface - are defined by their subsurface location, reflector curvature and dip. This general subsurface model produces stacking time surfaces which comprise the NMO/DMO, or Kirchhoff prestack depth migration time surfaces, as special cases. As a consequence, the CRS method generally provides a superior illumination of arbitrary reflectors, in comparison to NMO/DMO processing, or Kirchhoff PreSDM (Hubral, 1999).

Unlike the conventional NMO/DMO stacking surfaces, the CRS stacking surfaces extend beyond single common-midpoint (CMP) gathers. This implies a much higher fold, which leads to a better signal-to-noise ratio, and reflector continuity in the CRS stack. Moreover, dipping layers are strongly enhanced, since dip is explicity included in the CRS subsurface model.

The CRS stack is a technique to derive an optimum Zero Offset Stack. This is done by calculating the stacking operator that approximates the reflections in a multicoverage dataset best. The stacking operator depends on a couple of wavefield attributes, which are determined by coherency analysis. As results of this method not only the improved ZO Stack is calculated, but also the above mentioned wavefield attributes and coherency sections are delivered.

Figs. 1 and 2 illustrate the stacking surfaces for Migration to Zero Offset (MZO=NMO+DMO), Prestack Depth Migration (PreSDM) and CRS processing. The CRS stacking surface approximates the reflections much better and especially for the near offsets for far more traces in the vicinity of P0 than the MZO stacking surface. Its approximation of the reflections is nearly as good as is compared to the PreSDM stacking surface. The advantage of CRS stacking instead of PreSDM is that, these results could be obtained without knowing an exact velocity model in depth. This model is crucial for accurate PreSDM and can only be obtained through time consuming and costly processes.

fig. 1

fig. 2

crsTEEC Software and Technology

crsTEEC is a seismic processing software designed to support the Common Reflection Surface (CRS) techniques. This software package includes standard seismic processing methods (e.g. filters) and CRS specific algorithms like data driven CMP stack or data driven CRS stack.

crsTEEC can be used for a better ZO stack, for residual static corrections, for velocity inversion, or AVO data processing.

crsTEEC also includes modules for the computation of AVO attributes and for data visualization. Additional information from CRS processed data can be extracted from the calculated wavefield attributes. Using a new tomographic inversion process, these attributes are used for the calculation of a detailed interval velocity field. The inversion is very robust since it requires neither event picking in prestack data, nor continous events in poststack data.

Common-Reflection-Surface Stack using Kinematics Wavefield Attributes

The Common-Reflection-Surface (CRS) Stack (Hubral et al., 1999; Mann et al., 1999; Jäger et al., 2001) is a macro model independent stacking method to simulate a zero-offset section.

In order to generate a CRS stack, three attributes have to be found, viz., one emergence angle and two radii of wavefront curvatures. These attributes are related to characteristic properties of wave propagation.

CRS subsurface model

CRS reflection elements characterized by: location dip curvature For comparison: NMO stack NMO reflection points characterized by: location

CRS based Grid Tomography

CRS-Tomography is a new inversion method based on second-order traveltime approximations to the tomographic determination. The method makes use of traveltime information in the form of the kinematic wavefield attributes related to hypothetical NIP waves obtained from the seismic prestack data with the CRS stack. In order to extract the required attributes from the CRS stack results, it is sufficient to perform picking of reflection events in the stacked, simulated zero-offset section/volume, in which events are much easier to identify than in the prestack data. Because of the special model parametrization used in the tomographic inversion, which is similar to that of stereotomography (Billette and Lambaré, 1998), pick locations in the simulated zero-offset section/volume can be considered independently of each other. They do not need to follow interpreted horizons, but may be located on reflection events that are only locally coherent. A relatively small number of picks are sufficient to perform the tomographic inversion. This significantly simplifies and speeds up the picking process and allows to obtain a velocity model even in situations in which it is not possible, due to a low S/N ratio or complex reflector structure, to identify reflection events continously across the seismic section/volume.

• CRS tomography - a fast and efficient way to create a velocity-depth model
• Grid tomography on CRS attributes at selected locations only (poststack process)
• No Geologic interpretation is needed
• Works even on low fold data for which PreSDM fails
• Resulting smooth velocity model is ideal for migrations
• Cuts down turn-around time in PreSDM dramatically as model is about 95% correct

Example of CRS-Tomography in overthrust geology

Down is another example of the effort from CRS-Tomography in an Area with complicated Topography (great differences in the elevations). The CRS-Processing provide much better imaging of steeply dipping events and overall higher signal-to-noise ratio.

In the 2D case, the CRS attributes at certain, arbitrarily distributed grid points on the section are considered for the inversion process. Picking of the desired positions is carried out on the CRS stack itself. As this approach is not layer based only very few picks are required. They could for example be picked in zones with properly defined reflections whereas zones without reflection could be discarded. From the picked data points a grid model is created which serves as input to CRS attribute inversion. The result is a smooth velocity-depth model ideal for poststack depth migration or as initial model for prestack depth migration. Fig. 3 shows the results of a data example of the GOM with Salt.

	Fig. 3. Left: Original Result of the PreSTM. Right: Result of the CRS-PostSDM using Velocity field from CRS-Tomography
	Fig. 4. Left: Velocity field after Dix transformation of RMS velocities. Right: Velocity Field from CRS-Tomography
	Fig. 5. Details of both sections

CRS tomography and seismic acoustic inversion

Workflow to increase seismic Resolution of thin Reservoirs

Perform CRS based Grid Tomography:

• The result is a smooth velocity field
• Smooth Velocities derived from Sonic logs
• Compare smoothed sonic log velocities with result of grid based Tomography
• The sonic log derived low pass filtered velocities represent the low frequency trend of the tomography at well location
• Derive a seismic wavelet from the dataset
• Perform acoustic impedance inversion: density is taken from Gardner equation
• Check inversion result with original unfiltered sonic logs
• Identify the reservoirs and produce a horizon map of acoustic impedance at reservoir level

Low Frequency Model from CRS	Acoustic Impedance

CRS in AVO analysis

Flowchart of the method is applied to the multicoverage data to obtain the CRS attributes, that are subsequent used as input for the CRS inversion.

The obtained velocity model is then used by the kinematic migration algorithm. After migratin all common offset sections of the multicoverage data and stacking them to build a kinematic image, we can easily selected interested points to analyse, by means of AVO/AVA curves. Using the information in the geometrical-spreading factor tables, namely traveltimes and reflection point positions, the CRP gathers are extracted from the multicoverage data. Finally, picking the interested amplitude, the geometrical-spreading correction is applied to build up the AVO/AVA curves.

CRS-Gather: Traces in Pre-Stack domain of sevreal CDP's (Super Gather) considering reflector curvature Expected advantages of CRS-AVO: Reflector curvature already calculated High resolution velocity field Higher signal to noise ratio due to increased fold

CRS and advanced Processing Techniques for Reservoir characterization

The mentioned features of the CRS technique are not only interesting in terms of stacking, but also in terms of AVO analysis and seismic inversion. One key problem in AVO analysis is the observation, that NMO corrected reflection events are not exactly aligned along a constant traveltime. A residual moveout of the hyperbola still exists. Generally this is corrected by removing the residual moveout manually. Due to the non-hyperbolic velocity estimation this problem does not occur by using the CRS method. A comparison of the AVO attribute Gradient x Amplitude is presented is presented in Fig. 6 for the depth interval of the reservoir. Due to the improved velocity estimation and the resulting increase in signal/noise ratio by the CRS method, the AVO anomaly (red positive values) of the gasbearing reservoir now becomes much stronger. In other places artificial AVO-anomalies were observed resulting from a poor signal to noise ratio. Using the CRS-technique this problem could be better suppressed.

Due to the general increase in resolution found in CRS-processed sections also methods such as seismic inversion would certainly benefit from this methodology.

In conclusion the CRS technique has shown to be able to increase the seismic data quality which is one of the preconditions for a reliable reservoir characterization.