Seismotech, Ltd. will present reports at the 8th International Geological and Geophysical Conference and Exhibition "GeoEurasia-2025. Geological Exploration Technologies - Science and Business" on May 14-16, 2025, in Russian State Geological Exploration University by Sergo Ordzhonikidze (MGRI), located at bld. 23, Miklukho-Maklaya Street, Moscow
1. Application of wave field modelling in seismic data processing.
Roundtable “Modelling in seismic acquisition: current tasks and challenges”
The speaker: Ovchinnikov Kirill, Leading geophysicist, Seismotech Ltd.
The authors: Ovchinnikov Kirill, Silaenkov Oleg, Shalashnikov Andrey
Wavefield modelling is traditionally one of the fundamental tasks of seismic acquisition. Existing approaches vary from ray modelling and integral transforms up to numerical grid modelling methods based on finite difference or elemental methods.
Grid methods provide high accuracy and a comprehensive description of wave effects, but their computational cost makes them impractical for production processing. Radial modelling, even though it is successfully used for structural models building and observing systems planning , is not capable to transfer the dynamic parameters correctly and to display wave fields for complex and detailed described geological environments.
Under these conditions, the demigration technique based on the inverse migration transformation appears to be the most promising solution for industrial processing. The main idea of demigration is to use the estimates of reflection coefficients obtained during migration as a reflective characteristic of the medium (‘Reflectivity’), and to apply the same Green's functions to solve the direct problem as in depth migration. By combining the demigration technique and the layer-by-layer recomputation technique implemented in Prime, it is possible to quickly and accurately reconstruct model wavefields that reproduce only useful reflections. It is important that this technique allows the modelling of waves of a given type (reflected P-waves or S-waves, exchange waves, multiples, etc.), i.e., it complements full-wave modelling techniques that are otherwise difficult for interpretation.
These model gathers, possessing a given AVA/AVO dependence, can serve as a benchmark to evaluate and compensate for dynamic distortions arising during the processing of real data and to correct high-frequency statics. The formation of a reliable reference allows to increase significantly the spatial and depth resolution of migrated images, which directly affects the quality of interpretation of geological structures.
Thus, the application of demigration in production processing is justified as an optimal solution that allows to eliminate the limitations of both grid and ray methods. This approach provides the necessary accuracy and correct reproduction of the dynamics of wave fields, while the computational costs are significantly lower - the reduction reaches several orders of magnitude in comparison with traditional grid methods.
2. Model-based seismic data processing.
Roundtable “Depth imaging in seismic acquisition”
The speaker: Ovchinnikov Kirill, Leading geophysicist, Seismotech Ltd.
The authors: Anisimov Ruslan, Ovchinnikov Kirill, Silaenkov Oleg
The paper presents the concept of model-oriented seismic data processing, reflecting new approaches to complex processing and interpretation tasks solutions. Modern methods of seismic data processing have traditionally been based on simplified descriptions of time fields, but the transition to depth processing with an emphasis on dynamic modelling of wave fields makes it possible to take into account the physical properties of real geological media much more accurately.
In the proposed methodology, data processing is organised around more accurate models of the environment, which provides the creation of more meaningful transformations and the building of detailed depth images. This methodology is aimed at inverse problem solving that allows the recovery of wavefield parameters essential for interpretative conclusions.
The described approaches have been successfully tested on industrial projects using both synthetic and real seismic data. It is necessary to note that for the full work in the considered paradigm, the processing system should be equipped with a number of software and algorithmic innovations. The authors emphasise the need for close cooperation between research centres, leading industry enterprises and major oil companies to develop methods and optimise technological solutions.
In addition, the transition to model-oriented data processing facilitates the better comprehension of the dynamics of wave propagation in subsurface medium and opens new perspectives in complex analysis and interpretation of seismic signals. The development of this concept enables the creation of competitive software products that can meet the demands of the modern seismic market.
3. Q-tomography and Q-migration algorithms
Roundtable “Depth imaging in seismic acquisition”
The speaker: Anisimov Ruslan, Leading Software Developer, Seismotech Ltd.
The authors are: Anisimov Ruslan, Finikov Dmitry, Ovchinnikov Kirill, Shalashnikov Andrey
Today, Q-tomography and Q-migration technologies are among the most advanced in the field of analysing and correcting of the frequency-dependent absorption the effects. There are many ways of their implementation, but it is difficult to say positively that this problem is completely solved. Q-tomography and Q-migration methods widely known in practice have a number of significant algorithmic disadvantages, so this topic requires further investigation. This paper is devoted to the review of author's algorithms on this topic. The proposed methods include new approaches to the analysis of frequency-dependent absorption and new Q-migration methods that are free of the disadvantages of classical methods using inverse Q-filtering directly in the process of depth migration. The results of numerical modelling demonstrating the efficiency of the developed algorithms are provided.
4. A new EI-based algorithm for first breaks picking.
Petroleum geology and geophysics session
The speaker & the author: Anisimov Ruslan, Leading Software Developer, Seismotech Ltd.
It is difficult to overestimate the importance of the task of the Upper Part of the Section model (UPS) building and considering in seismic data processing. Considerable attention in this field has been paid to algorithms for the inverse problem solution, including various methods of inversion and first arrival tomography. However, the quality of any such algorithms is ultimately determined by the input data, i.e., the times of the first arrivals.
Although picking of the first arrivals has become a routine procedure years ago, it still remains a labour intensive process. Because the traditional algorithms and tools are often insufficiently robust to noise, complex wavefields and amplitude variations, most of the specialists' efforts are concentrated on rejecting and editing the obtained peaks. These difficulties are most evident in the processing of vibrodata. The main problem is that most existing algorithms are single-channel. Although many practical tools somehow use multichannel analysis at the post-processing stage, it does not solve the problem fundamentally.
Recently, methods based on artificial intelligence have been actively developed, but we want to show that the potential of classical approaches is far from being exhausted. This paper introduces a new multi-channel algorithm that redefines the approach to utilise classical single-channel metrics used in first-arrival picking. The proposed method significantly improves the robustness of the procedure to noisy data and reduces the labour cost of performing the subsequent editing of the results.
5. AGC and deconvolution. About waveform distortions during nonlinear transformations
Petroleum geology and geophysics session
The speaker: Kireeva Alina, geophysicist, Seismotech Ltd.
The authors are: Kireeva Alina, Finikov Dmitry
Statistical models and, in particular, the statistical seismic trace model are often used in seismic acquisition. In this case, it is usually assumed that, with the accuracy of smooth functions that compensate for the effect of geometric divergence, the trace amplitudes are stationary on the time scale. This is especially important when dealing with statistical inverse filtering (deconvolution) techniques. This property is often poorly fulfilled on the required sample volumes. Amplitude correction procedures, of course, allow ensuring the stationarity of records, but are considered incorrect for two reasons: 1) the recorded amplitudes carry important information about the reflective properties of the medium and 2) the application of a nonlinear adaptive procedure can lead to significant distortions of the convolution model of the seismic trace. The first reason can hardly be considered important, since the input of amplitude multipliers is easily reversible, while the second reason is analyzed in this paper. It is demonstrated experimentally that in the statistical sense (i.e., assuming random fluctuations of the pulse shape) the application of AGC (Automatic Gain Control) with reasonable parameters does not distort the convolutional model much and can be useful for the realization of statistical deconvolution algorithms.
6. F-XY seismic data filtering
Petroleum geology and geophysics session
The speaker: Finikov Dmitry, the Head of R&D Department, Seismotech Ltd.
The authors are: Anisimov Ruslan, Kireeva Alina, Ovchinnikov Kirill, Finikov Dmitry
We consider a popular class of filtering algorithms in seismic data processing which are intended for suppression of noise uncorrelated by spatial coordinates. These multichannel filters are calculated in the (f, x, y) domain, where f is frequency and x and y are spatial coordinates. In this paper we propose modifications of the algorithms for irregular observations processing and compensation of amplitude distortions of useful signals. These modifications eliminate the main weak points of the algorithms that limit the scope of their application at various processing stages. The application of machine learning methods for tuning the directional characteristics of the resulting filtering operators is discussed. Examples are provided to demonstrate how these methods achieve effective interference suppression and wavefield interpolation. Additionally, various generalizations of the F-XY filtering approach are described, enabling solutions for both area (3D) and line (2D) seismic data. These proposed algorithms have been implemented in software that allows big data processing ensuring its high speed and efficiency. The ability to use the algorithm for the processing of irregular 3D observations allows us to hope for the widest possible usage of F-XY filtering procedures to solve a variety of problems at different processing stages.
