IX Annual International Conference and Exhibition - Galperin Readings 2009

The problem of qualitative selection of useful waves at presence in the data essential regular and irregular noises is one of the basic problems in high definition seismic data processing. Use of method of projection in the valid parameters domains at selection increases a noise stability and detail of selection results. Methodical questions and features of realization of wave selection based on application of a priori model of medium for VSP data containing a different kind of frequency noises are considered in the paper. The example of method application in selection of offset shot point VSP data is given.

Migration of seismic data may be presented as two different steps – continuation of wavefield inside the medium and imaging in each point of medium. Wavefield continuation by finite-difference scheme is applicable for any complexity velocity models and is used in HDS (High Definition Seismic) technology. The most general uncompromising approach includes continuation of full 3C wavefields and vector wave selection in each interior point of the medium with solution of the inverse dynamic problem. The paper presents the results of a simple approach to migration of 1C data. In this case continuation of a wavefield of the reflected waves is made in reverse-time with correlation, for example, with the Ricker impulse on the DP-wave arrival time, calculated by the ray method, in all finite-difference knots. The results of such migration for 2D and 3D velocity models show the possibility of constructing realistic images for variable dip boundaries.

The paper represents sure method of reflected waves kinematics parameters estimation for surface seismic. This method can be used for velocity analysis in HDS technology. Universal model for reflected waves hodograph construction is produced. The results of kinematics analysis of reflections for real data are shown. Significant application of this method is the source data for inverse kinematics problem solution obtaining.

In the presentation the key points of the Gaussian Beam Summation Method (GBSM) and its applications to problems of exploration geophysics related with modeling of wave fields, depth migration and tomography are discussed. The efficiency of the proposed methods is demonstrated on the synthetic benchmark datasets such as: Salt Bag, BP Velocity and Sigsbee2A.

Artificial neural nets (ANN) are the self-learning, adaptive mathematical models. ANN successfully used in different areas to solve the classification problems, for data clustering, for regression task and many others. This article represents historical aspects and theoretical background of the neural nets. The main attention is focused on the innovation character of the neural nets, the ideas' content of the neural models, main stages of its construction and using. There is a list of advantages and disadvantages between neural nets and other, more traditional methods of solving approximation and regression tasks. Special analytical models demonstrated important regularities and features of the neural net paradigm. There were represented close relationships between neural nets theory and statistical methods. The real data was used to build an example of the creating, learning and using of the two neural nets classes - feed-forward neural net (MLP ) and general regression neural net (GRNN). This is an example to approximate complex relationships between seismic and well data and geological, geophysical and collectors' properties in the space between wells. There were shown possible ways to improve the accuracy of the neural net approximation. Also there was represented one of the method to value the confidence of the neural net model approximation.

The sourcewise approximation used in problems of gravity prospecting and prospecting. The present research devote adaptation the sourcewise approximation to problems of the seismic prospecting and geoelectrics.

As a rule, hydrocarbon pools in carbonate deposits are confined to nonanticline traps. Development of reservoir rocks in them are due to the secondary processes: leaching, solution and others. The difference in porosity between reservoir rocks and non-reservoir rocks is very small – to percent portions. In reservoir forecasting, such differences must be recognized on the attributes of a seismic signal (amplitude, energy and others). For this purpose it is necessary to have a detailed idea of rock petrophysical parameters (porosity, mineral composition and others). A geoacoustic model, linking the attributes of a seismic signal with petrophysical parameters is developed for establishing correlativity between them. The stages of seismic-geological modeling, which includes the development of petrophysical, geoacoustic and seismic-geological models, are shown on practical examples. It is shown that increasing rock capacity properties and density results in decreasing the velocity of advancing elastic fluctuations, acoustic rigidity and reflection coefficients. It presents a physical basis and a physical prerequisite for forecasting the development of reservoir rocks in a carbonate section by the character of changing the attributes of a seismic signal.

Examples of application of the VSP are giken, which illustrate possibilities of obtining the information about rock inaccessible by the instrumentality of exploration seismology and LOG.

The processing and interpretation of seismic data over the reservoirs was executed. Essential change of layers thickness of productive deposites was detection using spectrum and weivlet analyses. The positions of oil water contact and faults also was definition.