Luis Canales

Rapid VSP-CDP mapping of 3-D VSP data

Vertical seismic profiling–common depth point (VSP-CDP) mapping with rapid ray tracing in a horizontally layered velocity model is used to create 3-D image volumes using Blackfoot and Oseberg 3-D vertical seismic profiling (VSP) data. The ray-tracing algorithm uses Fermats principle and is specially programmed for the layered model. The algorithm is about ten times faster than either a 3-D VSP-CDP mapping program with an eikonal traveltime computation method or a 3-D VSP Kirchhoff migration program. The mapping method automatically separates the image zone from the nonimage zone within the 3-D image volume. The Oseberg data example shows that the lateral extent of the image zone created by the 3D VSP-CDP mapping is larger than that created by 3-D VSP Kirchhoff migration. The same sample result also provides high-frequency events at target zones. We include an analysis of the imaging error induced from using a horizontally layered model for the Oseberg data, indicating that the method is reliable in the presence of gently dipping structure.

Seismic velocity model building: CE in Dallas, 2 November

Depth conversion and depth imaging (depth migration) of 3-D seismic data require a geologically and geophysically competent velocity model. The constantly improving quality of seismic data and increasing survey accuracy demand a correspondingly high quality and accuracy standard for velocity models. In depth conversion of time‐based maps, for example, using a single regional velocity function is rapidly being replaced by depth conversion of both the interpreted surfaces and the seismic data using a single unified high quality 3-D velocity model.

Forward modeling attribute analysis for AVO and prestack depth migration

The attributes generated from forward modeling provide important information for acquisition design. Evaluating these attributes, such as amplitude, hit count, incident angles and others, can determine better shooting geometries. Because we illuminate the reflector by implementing wavefront construction based on multivalued, two-point raytracing, we are able to provide the raypath information for all receivers. The construction of the wavefront and raypath generates attributes for both the recording and the target horizon. Among these attributes, the aperture length is important for prestack depth migration and the incident angle is important for AVO analysis. These attributes combined with density information can be used to quantify the sensitivity of AVO to uncertainties in the rock properties (Sengupta and Rai,1998). Acquisition design based on common reflection stack (Chang et al., 2001) is an efficient way to optimize the shooting geometry. However, it will not provide good quality seismic data unless extra attributes are incorporated. In this paper, we will focus on attribute analysis for offset design, which is important for both AVO analysis and prestack depth migration. A synthetic model is used to present the concept of forward modeling illumination and our techniques are demonstrated on a realistic and complex salt model from the Gulf of Mexico.

Controlled stacking for improved quality of prestack depth‐migration results

Depth migration of seismic data is becoming more routine for imaging of complex geological structures. However, in some cases, the quality of the resulting depth section is lower than expected. A typical case is in imaging of salt bodies. The seismic images of steeply dipping salt flanks are frequently broken, smeared, and difficult to interpret. The study presented in this paper demonstrates one source of this imaging problem, and suggests a solution we call controlled stacking. The core of the solution is the correct muting of partial images produced by prestack depth migration (preSDM) before sorting and stacking to obtain the final depth section.

Mixed‐grid solution of the 3‐D Eikonal equation

Calculation of 3D traveltimes is needed for the application of Kirchhoff integral based prestack depth migration. 3D prestack imaging requires a significantly large amount of computation, where a major part is dedicated to traveltime calculations. Therefore, a practical method of calculating traveltimes is needed. Our algorithm was developed with the following objectives: (a) the numerical scheme should be simple and efficient, (b) the traveltimes should represent arrival times of body waves, and (c) the calculated traveltime field should be accurate for steep dip structures. To achieve these objectives, we (a) directly solve the 3D Eikonal equation on a Cartesian coordinate system, (b) modify the over-critical arrivals on a special cylindrical coordinate system, and (c) apply a spatial convolutional operator to overcome errors introduced by low order approximations.

On Accurate Imaging of Steep Salt Flanks

The accurate imaging of steep flanks of salt domes has long been a challenging problem for exploration geophysicists. The problem continues to be relevant today as more and more salt domes are being shot with 3D seismic in order to exploit remaining hydrocarbon reserves. Unfortunately, the seismic images of steep salt flanks are often not satisfactory; they are either broken or appear as smeared events. Interpretation of the salt body in these cases is difficult and somewhat confusing (see figure la).

VSP Mapping Error For Dipping Horizons Using A Horizontally Layered Model

INTRODUCTION Among all the velocity models, the horizontally layered model is probably the most useful one in the early stages of model building. One reason for its frequent use is that predominantly, the earth varies vertically. The other reason is that ray tracing is easy to handle in this structure. The use of a horizontally layered model may be even more applicable for VSP surveys because the survey area is close to a well (Biquart, 1998; Zhang et al., 1997). However, as long as the structure of interest is not horizontal, error estimation is needed to provide an accurate image of the subsurface. For example, Chen and Peron used a horizontally layered model for 3D VSP-CDP mapping of Oseberg VSP data (Chen and Peron, 1998) and the image showed some mildly dipping structure. It was not obvious at that time how much error would be involved in the image through the use of the horizontally layered model. This paper gives an estimation of the VSP CDP mapping error for the Oseberg data example given by Chen and Peron (1998). A horizon-mapping method based on a ray-tracing algorithm is used. For simplicity, a 2D environment with the section parallel to maximum dipping direction is chosen. It is not difficult to extend the theory to a 3D case.

3D traveltime calculation based on Fermat's principle

SUMMARY We present an algorithm for the fast and accurate computation of the multi-valued traveltime and amplitude in a complex model. It has been designed for 3D prestack Kirchhoff based subsurface imaging. Sampling of wavefield is performed along rays by the wavefront construction. Traveltime and amplitude tables are calculated based on the Fermat’s principle. The main contribution of this paper is the efficient way to implement Fermat’s principle by the recursive subdivision on several loops of the ray cells [10]. Traveltime table calculation by this method is not only efficient but also provides an accurate results on complex geological models with the size of grids in the 3D model compared to the traditional search and volumetric interpolation [1][11].

Stable wavelet estimation

Estimation and removal of the source wavelet prior to stacking is very important when the source wavelet is variable. Unfortunately, this very often coincides with noisy areas, making the wavelet estimation unreliable. In our development we consider the use of some kind of ensemble averaging since single trace deconvolution is unsatisfactory. We also find it necessary to look for areas with good signal to noise ratio “S/N” that avoid both random and coherent noise trends. This in turn forces us to consider the use of short windows, hoping that the statistical resolution lost by using short windows can be regained by use of ensemble averaging.