The paper discusses exploration and development issues using artificial intelligence methods based on new seismic attributes of the RTH (Reverse Time Holography) method and well drilling data. RTH attributes are based on two-stage seismic data processing: on decomposition the initial common shot gathers in common image gathers, using the time-reversal algorithms and on synthesis a seismic attributes. It is shown that a detailed analysis of the joint behavior of two vectors: the velocity vector in forward wave and the velocity vector in time-reversed backward scattering wave provides detailed information about the medium. The main differences between RTH attributes and traditional ones obtained during migration are their voxel nature and hyperattributivity. It turned out that this is a key advantage of the new approach to solving problems of geological prediction using artificial intelligence methods. The paper presents the results of using the new method for processing and interpreting 3D seismic data, as well as geological prediction based on RTH attributes for a number of oil and gas fields.
The problems of predicting petrophysical parameters, as well as flow rates and other production characteristics are of significant interest, both in theoretical and practical aspects in the oil and gas industry. The inputs to the prediction are seismic attributes and well-log data. Presently seismic attributes are typically calclated in the time domain, while well-log data are depth-specific. This, along with the different detail of the information obtained in seismic and in the well, is the main difficulty in integrating seismic data and well-log data for the purpose of geological prediction of the properties of the medium throughout the entire space. Therefore, in most previous studies, well-log data are recalculated into the time domain, where their integration is carried out. The fundamental feature of the approach described in this paper, from all previous ones, is the use of new generation depth-based seismic attributes obtained using the seismic processing method that implements principles of seismic holography and wave front reversal in time - the RTH (Reverse Time Holography) method (Erokhin, 2019). The method is vector extensions of the well-known method of depth migration based on wave reversal in time - RTM (Reverse Time Migration) (Baysal et.al., 1983; Whitmore, 1983, McMechan, 1983). The RTH method includes, as a special case, method based on a common image point - Angle Domain RTM (Yoon and Marfurd, 2006; Alkhalifah, 2015), diffraction analysis method ES360 (Koren and Ravve, 2011), CSP (Kremlev et. al., 2011), method of angular anisotropy of reflection - Amplitude versus Offset (AVO) (Chopra and Castagna, 2014), acoustic inversion method (Tarantola, 1984), velocity tomography method based on full-wave inversion (Virieux and Operto, 2009) or based on beam tomography (Popovici et.al., 2016). RTH is a voxel-based method, that is, the assessment of seismic attributes is carried out in each cell (voxel) of geological space independently of each other. Voxels are of arbitrary size, and their coordinates are fixed in the space they fill. The set of seismic RTH attributes includes, in addition to all known attributes, a number of previously unknown ones. The total number of seismic attributes obtained based on the parameters estimation of multidimensional (10-dimensional) statistical distribution in the RTH method reaches several hundred (Erokhin, 2022).
