Matthias G. Imhof

Diffraction enhancement in prestack seismic data

Seismic diffractions are often considered noise and are intentionally or implicitly suppressed during processing. Diffraction-like events include true diffractions, wave conversions, or fracture waves which may contain valuable information about the subsurface and could be used for interpretation or imaging. Using synthetic and field data, we examine workflows to separate diffractions from reflections that allow enhancement of diffraction-like signals and suppression of reflections. The workflows consist of combinations of standard processing modules. Most workflows apply normal moveout corrections to flatten reflection hyperbolas, which eases their removal. We observe that the most effective techniques are the decomposition of seismic gathers into eigensections and flows based on Radon transformations.

Amplitude preservation of Radon-based multiple-removal filters

This study examines the effect of filtering in the Radon transform domain on reflection amplitudes. Radon filters are often used for removal of multiple reflections from normal moveout-corrected seismic data. The unweighted solution to the Radon transform reduces reflection amplitudes at both near and far offsets due to a truncation effect. However, the weighted solutions to the transform produce localized events in the transform domain, which minimizes this truncation effect. Synthetic examples suggest that filters designed in the Radon domain based on a weighted solution to the linear, parabolic, or hyperbolic transforms preserve the near- and far-offset reflection amplitudes while removing the multiples; whereas the unweighted solutions diminish reflection amplitudes which may distort subsequent amplitude-versus-offset (AVO) analysis.

Scale dependence of reflection and transmission coefficients

Well logs show that heterogeneities occur at many different depth scales. This study examines the effects of these heterogeneities on the propagation of seismic waves, and specifically the dependence of reflection and transmission on the spatial scale content of the medium. Wavelet transformations are used to filter certain spatial scales from an acoustic sonic log. The scale‐filtered logs are used to construct layerstack models for which reflection and transmission seismograms are computed. The modified logs are also used to calculate frequency dependent reflection and transmission coefficients as functions of scale content. It is observed that features shorter than one‐fourth of the dominant wavelength have little effect on the reflection and transmission of seismic waves. Features larger than the dominant wavelength affect arrival times of individual packets within the wavetrain, but often these features hardly alter the overall appearance of individual wave packets. Reflection and transmission coda ar...

Seismostratigraphic inversion: Appraisal, ambiguity, and uncertainty

Geologic process models predict the geometry of geologic strata and their petrophysical properties, based on mathematical models of geological processes that affect the formation and evolution of geologic strata. Such processes include erosion, sediment transport, and deposition. The resulting forward model is typically nonlinear. Given observations and a misfit measure, one may attempt inversion of these models to estimate process parameters that yield compatible predictions. For seismostratigraphic inversion, seismic data are used as observations. We tested such an algorithm in a prograding-delta environment to examine the effect of using different seismic attributes as observations and, thus, different choices of misfit measures. The first measure, based on the degree of parallelism between seismic reflectors and modeled geologic strata, demonstrated a trade-off between geologic time and the sediment-influx rate used to parameterize the model. A second misfit measure used unwrapped seismic instantaneous phase as a crude proxy to relative geologic time, which regularized the model parameters. Then last, we combined the two measures to take advantage of their individual characteristics. For most of these inversion experiments, we obtained results that capture the geometry of the geologic strata as observed on the seismic data. With the exception of the depositional time-rate trade-off, where the same strata can be obtained in a shorter geologic interval when rates are increased, we found the inversion to be surprisingly stable, with a unique cluster of acceptable parameters, despite the nonlinearity of the geologic forward model.

Analysis and application of coal-seam seismic waves for detecting abandoned mines

Two in-seam reflection surveys and one transmission survey were acquired at an abandoned underground mine near Hurley, Virginia, to demonstrate the feasibility of detecting abandoned-mine voids utilizing coal-seam seismic waves. Standard, commonly available tools for seismic reflection processing were used. The mine was detected and located by using trapped coal-seam seismic waves observed in both the transmission and reflection data. Detecting the void, however, was not good enough to replace drilling entirely. We conclude that in-seam seismic methods can be used for detection; but if a potential void is detected, focused drilling should be applied for accurate mapping and to circumvent potentially hazardous areas.

Modeling offset-dependent reflectivity for time-lapse monitoring of water-flood production in thin-layered reservoirs

The objective of this case study is to predict whether 5 years of water-flood production from a thinly layered Gulf of Mexico reservoir will change its seismic amplitude-variation-with-offset (AVO) response in a detectable manner. Density and velocity profiles were computed from in situ wireline logs for 100% oil, gas, and brine saturations and for a 5-year prediction that was based on a fluid-flow and production simulation. Analytical AVO curves for simple half-space models did not match AVO curves extracted from synthetic seismograms computed with a full-waveform layer-stack algorithm. Several different amplitude corrections were tried to reduce the AVO curves from the synthetic data to the analytical ones, but, ultimately, none was deemed satisfactory. Instead, AVO change attributes based on relative changes, polarity changes, or ratios were used. Attributes based on the change of AVO gradient were perceived to be most diagnostic of the water flood, but they were also overly sensitive to interference noise and amplitude correction errors. For field data from the study area, a large decrease in intercept magnitude may be the best indicator of the waterfront.

Data Consistent Phase Unwrapping in Three Dimensions

Stark (2003) demonstrated that in siliclastic sequences, unwrapped instantaneous phase may be used as a proxy for relative geologic time. Unwrapping phase trace by trace, however, can generate artifacts which reduce usefulness and interpretability of the resulting relative-age volumes. Hence, I propose use of a multidimensional, data-consistent algorithm which unwraps all traces of seismic sections or volumes simultaneously.

Calculating the Seismic Effect of 3-D Underground Structures and Topography with the Finite-Difference Method

I present a finite-difference algorithm to calculate the seismic response of cavities or underground structures in a complex 3-D medium. Topography in three dimensions is just a special case, and hence, is automatically included. The algorithm is based on the imaging method proposed by Levander (1988) and Robertsson (1996). I observed that traditional absorbing boundary conditions were either ineffective or instable. To circumvent this problem, I coupled the imaging technique with perfectly matched layers (PML). For the sake of simplicity, however, not only a boundary layer, but the entire computation is performed for a perfectly matched medium.

Poststack interpretive static correction

The paper describes a novel technique to interpret and map unconformity on 3D seismic data with misties in a very complex geological setting. The study area, Coalinga giant oil field in California, represents a dynamic depositional setting along with depositional hiatuses in a relatively unconsolidated, shallow Miocene clastic reservoir. The flow distribution of the field is mainly governed by the presence of three distinct unconformities and shaliness. The available, merged post-stack 3D datasets were acquired over different time periods in a field with numerous steam injection activities, which caused misties and reflector discontinuities on the seismic sections. Line tying of these unconformity reflectors is problematic. Reverting to reprocessing the data is time consuming and uneconomical; a novel technique is attempted to aid our interpretation. We map a deeper, relatively continuous reflector and flatten to with some time reference. The misties are reduced considerably on this flattened volume and the continuity of these unconformities is easily observed. The well data are posted on this volume with consideration of this flattening time. We then interpret these unconformities in the original seismic volume with the aid of this flattened volume.

Resolving Small Objects Using Seismic Traveltime Tomography

It is often claimed that the spatial resolution limit for traveltime tomography is the first Fresnel zone associated with the dominant frequency in the data. To test this assertion, synthetic seismic data were generated for traveltime picking and inversion for a single, small velocity anomaly embedded in a homogeneous background velocity. All picking methods produced accurate raytheoretical (infinite-frequency) picks from noise-free seismic data for objects much smaller than the dominantfrequency Fresnel zone. All methods detected the presence of objects smaller than a wavelength. The inversion of these traveltime picks always recovered the position and shape of the object. Random noise was added to the synthetic seismic data. Pick times with different noise realizations were statistically centered on the noise-free pick, not the time that would be recorded in the absence of the object. Trace stacking prior to picking or the averaging of many picks improves the signal-to-noise ratio and can extract signal that is not resolved on an individual pick. An averaging of traveltime picks also occurs during tomographic inversion. This inherent signal-to-noise improvement allows tomography to image small objects that are not resolved in individual trace picks. The resolution of tomography is limited not by the Fresnel zone associated with the dominant frequency, but by the accuracy of the traveltime picks and the density of ray coverage.