EAGE/SEG Workshop - Depth Imaging of Reservoir Attributes

Seismic survey design is often a relatively qualitative procedure that relies on experience and economic constraints rather than quantitative predictions of the image quality produced by different acquisition geometries. Simple ray tracing procedures provide one technique for obtaining some quantitative input by measure binning along target interfaces.

3-D Prestack Depth Migration has greatly improved our ability to image reservoirs in complex areas, such as below salt and carbonates in the North Sea or beneath salt bodies in the Gulf of Mexico. But what effect does the overlying structures and abrupt velocity variations have on seismic attributes such as amplitude anomalies or amplitude variation with offset (AVO)?

Data acquired with the parallel geornetry can be described as a collection of common-offset gathers. Prestack migration of such data can be conveniently considered as the repeated application of the migration proces to all common-offset gathers.

The convergence of conventional reflection tomography is often uncertain when the starting velocity function is too far from the correct one. The time-domain reflection tomography that we present in this paper is more robust than conventional depthdomain reflection tomography.

The "flatness" of events in the Common Image Gathers (CIGs) used for AVO analysis, is directly related to the quality of the velocity model [4,6]. The method we present here for velocity estimation in 2D media is precisely adapted to further AVO analysis, as its aim is to flatten the depth migrated CIGs. For an automatic velocity estimation where no picking is introduced, the choice of the objective cost function which measures the "flatness" of the coherency panels, is crucial. We explore some proporties of the Differential Semblance function, introduced by [7], to prove its efficiency in 2D velocity estimation. Synthetic tests are presented here, and show that at least in these cases, total optimization algorithms may be applied.

Common image panels are commonly used for migration based velocity analysis and for migration based AVO analysis. The advantages of common offset image gather with respect to common shot image gathers are now well established [4] but also their poor quality in case of complex models [3]. The reasons of this failure may appear when the problem is analyzed in the frame of the asymptotic linearized inverse theory [7, 6]. We propose here an analysis of the relevance of common image gather analysis in case of velocity models with triplicated ray fields. We show that individual common shot or common offset may provide inconsistent images of the subsurface, and we propose a strategy for common anale imaging avoiding the former artefacts. Comparisons of common shot, common offset and common angle image gathers are presented for 2D complex Marmousi dataset.

The proposed reflector migration is a fast and accurate prestack depth migration designed for interval velocity and depth interface determination and for AVO analysis. The output of the migration consists of CRP gathers in windows centered at the reflecting horizons (CRP migrated panels) and the corresponding CRP ray paths for the given output offsets. Related approaches for 2D data were described in a patent by Johnson (1990), Landa and Sorin (1993) and Kosloffet et. al. (1995). This work is designed for 3D complex geological structures with a special treatment for obtaining true amplitude reflection events.

AVO migration/inversion has been developed for estimating AVO attributes for the case of moderately complex structures. There are several sources of uncertainty in this procedure, the largest of which is usually the estimation of a suitable depth migration velocity field.

Irregular and sparse sampling of the wavefield in 3-D surveys generates spatial aliasing. Wavefield interpolation yields densely sampled data either operating in a suitable domain where sampling is more regular and can then be densified (e.g., in common shot domain) or by exploiting the predictability of the wavefield as parameterized by one (or more) plane events. In this paper we present an analysis of the sensitivity to velocity of the 3-D shot continuation operator (3-D SCO) for interpolation of 3-D acquisition geometries and we compare the 3-D SCO interpolation with prediction-error-filter (PEF) approach.

Three basic processing methodologies for amplitude-versus-offset (AVO) / amplitude-versus- angle (AVA) studies are discussed and compared: time processing, time imaging and depth imaging.

Abstract was made available at the workshop

3-D ray + Born and 3-D ray + Kirchhoff migration/inversion schemes are considered as the most accurate tools for quantitative depth imaging in 3-D heterogeneous media. The both formulations are presented here with a brief theoretical review and some comparisons on a 3-D real data set. Important theoretical and practical points (such as acquisition footprints, velocity model quality, image post-processing) are underlined with the help of Common Offset gathers in order to anderstand the method capacities for subsequent reservoir attribute studies.

The limitations of classical ray theory, in case of triplicated ray fields, are well known singularities at caustics, shadow zones, ... Lots of theoretical works have been done for avoiding these artefacts.

Even when subsurface velocities vary only moderately, accurate imaging of reservoir attributes in depth remains a challenge. Two crucial components of the solution are the determination of an accurate velocity model and the use of an appropriate imaging algorithm. We examine the depth accuracy and qualitative AVO response of images generated using different velocity models and different imaging methods. We conclude that velocities determined from offset checkshot data in conjunction with a 3-D Kirchhoff imaging algorithm yield prestack images whose depths and amplitudes are in better agreement with well data than 3-D prestack migration implemented as a multi-step process (MZO cascaded with zero offset migration).

Here we present a technique for estimating the velocity associated with a final preSDM image in great detail. Essentially we employ techniques of continuons velocity estimation used for iterative model update, but apply them in a different context. Rather that using the velocity estimator strictly for velocity model building (on a set of sparse lines), we employ the estimation techniques to the ARP gathers resulting from the final migration for all CRPs in the 3D survey. This achieves two things: firstly it estimates a residual velocity correction for stacking the final CRP gathers to obtain an enhanced image, and secondly, it delivers a dense estimate of velocity for the entire 3D volume in a geologically coherent manner. It is this second feature which is the subject of this paper.

In the Vienna Basin are located important hydrocarbon reservoirs, characterised by fractured rocks in a complex fault system. The pre-stack depth migration is a basic tool for identifying the major geological formations. This method is particularly effective when the necessary macro-model for the velocity field and layer interfaces is provided by seismic tomography. There is a trade-off, However, from the practical point of view: seismic tomography is more necessary for relevant complexities but, in such cases, the picking of traveltimes becomes extremely difficult, if not unfeasible. So a goal of our work was to verify the effectiveness of the tomographic imaging in this difficult area.

We present the new data of the delineation of reservoirs and connected fault zones based on the simultaneous registration of the readings of a specially selected group of high precision gravimeters and measurement of natural microseisms.

Operto et al. (1997) presented an application of 3D ray+Born inversion to an onshore-type sub-dataset for imaging a dip section of the SEG/EAGE Overthrust model.

An erroneous amplitude picking can severely degrade AVO (Amplitude Versus Offset) analysis results and may lead to incorrect drilling decisions. A major cause of errors in AVO picking is the presence of coherent and incoherent noise that interfere with the target reflection. AVO picking is therefore a crucial step in AVO analysis and care must be taken in the rejection of coherent noise and in the exploitation of the available range of incidente angles. The joint estimation of AVO and kinematic parameters (JAK: Joint Amplitude and Kinematic) (Grion, Mazzotti, Spagnolini, 1998) is an AVO picking technique that meets these requirements.