Preface 1. Introduction to rock physics 2. Rock physics interpretation of texture, lithology and compaction 3. Statistical rock physics: combining rock physics, information theory, and statistics to reduce uncertainty 4. Common techniques for quantitative seismic interpretation 5. Case studies: lithology and pore-fluid prediction from seismic data 6. Workflows and guide lines 7. Hands-on References Index.

Quantitative Seismic Interpretation: Applying Rock Physics Tools to Reduce Interpretation Risk

Two-dimensional, fenced 2-D, and 3-D isosurface displays of some realistic 3-D seismic models built in the lower Miocene Powderhorn Field, Calhoun County, Texas, demonstrate that a seismic event does not necessarily follow an impedance boundary defined by a geological time surface. Instead, the position of a filtered impedance boundary relative to the geological time surface may vary with seismic frequency because of inadequate resolution of seismic data and to the en echelon or ramp arrangement of impedance anomalies of sandstone. Except for some relatively time-parallel seismic events, the correlation error of event picking is large enough to distort or even miss the majority of the target zone on stratal slices. In some cases, reflections from sandstone bodies in different depositional units interfere to form a single event and, in one instance, an event tying as many as six depositional units (interbedded sandy and shaly layers) over 50 m was observed. Frequency independence is a necessary condition for selecting time-parallel reference events. Instead of event picking, phantom mapping between such reference events is a better technique for picking stratal slices, making it possible to map detailed depositional facies within reservoir sequences routinely and reliably from 3-D seismic data.

Stratal slicing, Part I: Realistic 3-D seismic model

We introduce a seismic-sedimentological approach for mapping high-frequency (fourth-order) sequences and systems tracts using well and three-dimensional seismic data. Key techniques include (1) conditioning seismic data to log lithology by 90 phasing for better well-log integration and (2) imaging and interpreting the sequential, planoform geomorphology of the depositional systems. We recommend a new interpretation procedure that shifts the emphasis of high-frequency sequence-stratigraphic studies from interpreting vertical seismic sections to analyzing more horizontal, high-resolution, seismic-geomorphologic information.This case study shows that stratal slicing in lithology-conditioned seismic data provides sequential seismic imagery of generally contemporaneous depositional systems. This imagery, in turn, serves as a basis for recognizing and mapping high-frequency systems tracts, sequence boundaries, and sequences in a geologic-time domain. In Miocene strata of offshore Louisiana, fourth-order sequences or sequence sets from well data can be seismically mapped at a resolution equivalent to 30 ft (10 m) in thickness from a 30-Hz dominant-frequency seismic data set. The resolution is sufficient for an accurate reconstruction of the high-frequency sequence-stratigraphic framework in the region of seismic coverage outside well control.

High-frequency sequence stratigraphy from seismic sedimentology: Applied to Miocene, Vermilion Block 50, Tiger Shoal area, offshore Louisiana

Predictive painting is a numerical algorithm that spreads information in 3D volumes according to the local structure of seismic events. The algorithm consists of two steps. First, local spatially variable inline and crossline slopes of seismic events are estimated by the plane-wave-destruction method. Next, a seed trace is inserted in the volume, and the information contained in that trace is spread inside the volume, thus automatically “painting” the data space. Immediate applications of this technique include automatic horizon picking and flattening in applications to both prestack and poststack seismic data analysis. Synthetic and field data tests demonstrate the effectiveness of predictive painting.

Predictive painting of 3D seismic volumes

We present an efficient full-volume automatic dense-picking method for flattening seismic data. First local dips (stepouts) are calculated over the entire seismic volume. The dips are then resolved into time shifts (or depth shifts) using a nonlinear Gauss-Newton iterative approach that exploits fast Fourier transforms to minimize computation time. To handle faults (discontinuous reflections), we apply a weighted inversion scheme. The weight identifies locations of faults, allowing dips to be summed around the faults to reduce the influence of erroneous dip estimates near the fault. If a fault model is not provided, we can estimate a suitable weight (essentially a fault indicator) within our inversion using an iteratively reweighted least squares (IRLS) method. The method is tested successfully on both synthetic and field data sets of varying degrees of complexity, including salt piercements, angular unconformities, and laterally limited faults.

/pdf/flattening-without-picking-1u7kx7d9jt.pdf

Flattening without picking

We generated a set of three-dimensional (3-D) seismic models of a lower Miocene progradational microtidal shore-zone system at the Powderhorn field, Calhoun County, Texas. These models were based on detailed lithological mapping of 250 m of stratigraphic section over an area of 13 × 9 km. The mapping was guided by detailed facies analysis based on wireline logs from 115 wells. Fourteen sandy depositional units averaging 3-30 m in thickness are encased in 15 shale units. Density-neutron and spontaneous potential (SP) logs in five recently drilled wells were used to calculate effective porosity and shaliness, which correlate well with P-wave velocity and bulk density logs. Those properties were then extended to older wells lacking acoustic, density, and other porosity ogs through regression with SP and resistivity logs to build 3-D properties models, including models of transit time and impedance. The impedance model was mapped from x, y, z to x, y, t, where t is two-way vertical time. The impedance model was converted to primary reflection coefficient series that were then convolved with zero-phase wavelets of different frequencies to produce the seismic models. The results show how the depositional facies at reservoir scale might be illustrated by 3-D seismic data, and how the seismic resolution of depositional facies changes with seismic frequency, stratigraphic position, and facies patterns. The Powderhorn reservoir sequence is interpreted as a barrier bar/lagoon depositional system associated with coastal stream plains, bayhead deltas, and small wave-dominated deltas corresponding to frequent relative changes of sea level. To detect these types of deposits, the seismic frequency should be selected such that the seismic data are tuned to the maximum thickness to establish a linear relationship between seismic amplitude and facies, and to achieve the best signal-to-noise ratio. The facies associated with the environments where main sandstones are surrounded by muddy deposits, such as coastal stream plain, bayhead delta, delta plain, and backbarrier/lagoon transition, are well illustrated in our seismic model. Lack of muddy deposits, for example in a wave-dominated delta, c uld result in obscure facies boundaries and therefore unclear lithofacies patterns. In this study, poorly compacted clean sandstone and pure shale have similar impedances, and consequently produce similar seismic responses. The resulting ambiguity may be reduced by careful analysis of facies relationships if some well data are available, and probably by the amplitude vs. offset (AVO) method. The stratal resolution of seismic imagery depends not only on wavelet frequency, but also on the stratigraphic position of a given reservoir, or the magnitude of geological interference.

Facies Mapping from Three-Dimensional Seismic Data: Potential and Guidelines from a Tertiary Sandstone-Shale Sequence Model, Powderhorn Field, Calhoun County, Texas

We discuss some of the first observations based on the realistic 3-dimensional seismic models that we built in an accompanying paper. We illustrate the effect of hydrocarbons, using Gassman’s equation, and show that gas and light oil should dominate the stratal-slice amplitude displays as anomalous bright spots. The stratal time slices show that most reservoirs can be pictured and interpreted in terms of depositional facies if we select the appropriate wavelet frequency. The spatial resolution of seismic data depends not only on the wavelet frequency, but also on the stratigraphic position of a given reservoir. This is important in designing a reservoir-targeted 3-dimensional seismic survey. Further analysis of the seismic models demonstrates that in modem data having frequency components of 20 to 50 Hz, seismic events do not necessarily follow geological time lines. For sand-prone sequences like the one modeled here, seismic-event-peak picking is a questionable approach for high-frequency sequence stratigraphy and for stratal-slice displays.

Kenneth T. Barrow

Papers

Stratal slicing, Part I: Realistic 3-D seismic model

Facies Mapping from Three-Dimensional Seismic Data: Potential and Guidelines from a Tertiary Sandstone-Shale Sequence Model, Powderhorn Field, Calhoun County, Texas

Facies-Guided 3-D Seismic Modeling: Part 2, Model Interpretation