Dennis Cooke

Generalized linear inversion of reflection seismic data

Generalized linear inversion, sometimes known as model perturbation, nonlinear regression, or inverse modeling, is applied to synthetic and real seismic data sets with the objective of obtaining an impedance profile as a function of time. The impedances solved for are parameterized in a manner that describes the unknown earth using fewer variables than previous seismic generalized linear inversion techniques. In this application only single traces of common‐midpoint (CMP) processed data will be inverted. The method of generalized linear inversion (GLI) presented here is designed to improve on the shortcomings of recursive inversion with respect to relative and absolute scale of the impedance results, resolution of impedance boundaries, and distortion from residual wavelet effects. In obtaining these goals other advantageous aspects of GLI were discovered. For example, it is insensitive to noise in many cases, and it will allow an interpreter to fix the impedance of any number of known lithologies in an in...

Simultaneous time imaging, velocity estimation, and multiple suppression using local event slopes

Starting with the double-square-root equation we derive expressions for a velocity-independent prestack time migration and for the associated migration velocity. We then use that velocity to identify multiples and suppress them as part of the imaging step. To describe our algorithm, workflow, and products, we use the terms velocity-independent and oriented. While velocity-independent imaging does not require an input migration velocity, it does require input p -values (also called local event slopes) measured in both the shot and receiver domains. There are many possible methods of calculating these required input p -values, perhaps the simplest is to compute the ratio of instantaneous spatial frequency to instantaneous temporal frequency. Using a synthetic data set rich in multiples, we test the oriented algorithm and generate migrated prestack gathers, the oriented migration velocity field, and stacked migrations. We use oriented migration velocities for prestack multiple suppression. Without this multi...

Probabilistic Neural Network inversion for characterization of coalbed methane

Summary The seismic guided estimation of reservoir properties away from the wells is a common problem that geophysicists, geologists and reservoir engineers face every day. This problem is due to low resolution of seismic data as well as the lack of proper models that link the seismic data to the borehole data. Geostatistical methods help resolve this problem, but these methods rely only on a linear fit between seismic attributes and reservoir parameters. Artificial neural networks are the best method to relate the non-linear fit between the borehole parameters to seismic volume parameters to better understand the heterogeneity of the reservoir properties. Probabilistic neural networks (PNN) are used to invert the seismic data of coalbed methane (CBM) field from northeast Australia to better understand the reservoir properties of the coals sandwiched between sands and shales. PNN has not only helped to improve the vertical resolution but also the lateral variation in the heterogeneity of the reservoir.

Correlating heterogeneous production to seismic curvature attributes in an Australian coalbed methane field

Australia is the worlds second largest producer of coalbed methane (Faiz, 2008). Characterized as a marginal tectonic setting, the Bowen Basin of Queensland displays a thick succession of numerous thin bituminous rank coal seams. The areas heterogeneous production is particularly perplexing; it is not unusual for production to change as much as 50-75% between neighboring wells within a few kilometers of one another. A myriad of factors can affect coalbed methane production to include coal thickness, coal cleat architecture, local maximum horizontal stress direction, and the in situ stress magnitude.

Sensitivity of velocity-less imaging using horizontal slownesses in 3D

We consider signal propagation in one layer with a constant velocity-gradient: layer in which the signal velocity increases linearly with depth. We use the traveltime and the four derivatives of the traveltime with respect to the locations of the source a

Quantitative reservoir characterization at B field, Java Sea: A statistical approach and a model based approach

This is a seismic attribute case history that is notable for the follow reasons: • Quantitative attribute analysis is performed using two different approaches; a conventional statistical approach (valid for a field development with many wells) and a theoretical model-based approach that requires a minimum of one well. • With the statistical approach, a fluid (oil-versus-water) amplitude effect is observed. • Initially, the reservoirs’ synthetic amplitudes were consistently lower that the observed amplitudes. This discrepancy was explained only by looking at the fluid effects of oil, gas, water, and reservoir pressure drop caused by oil production. • Within the model-based approach, the effects of reservoir thickness, reservoir quality and reservoir fluid on seismic amplitude are presented on one graph and applied to the mapped amplitudes of undrilled prospects.