Processamento de dados sísmicos terrestres 3D: novas abordagens e soluções

Abstract

The processing of 3D land seismic data is always a great challenge for the petroleum industry, which constitutes a current, relevant, and open to contributions. Among the main problems in the processing flow, we highlight the static errors, obtaining the velocity field, and attenuation of the ground roll. As a product of this thesis, we developed new methodologies and algorithms that improve the 3D ground seismic processing flowchart and, consequently, the final seismic image. The first contribution is associated with the definition of picking the first break in 3D ground seismic data using the Dynamic Time Warping (DTW) algorithm. In this step, the DTW algorithm that uses the similarity between seismic traces, calculated from a cumulative cost matrix, identifies the first break in the seismogram based on a seed pre-defined by the analyst. We applied the method to the Polo-Miranga seismic cube in the Recôncavo Sedimentary Basin and Blackfoot (located in Alberta, Canada). The results were satisfactory for the entire range of offsets of the seismic data. In addition, we performed the statics calculations, and the results proved consistent as they attenuated the temporal distortions present in the seismic traces caused by the topographic and thickness variation of the low-velocity zone. The second contribution is associated with estimating seismic velocity profiles from the decomposition of slowness profiles using the Exponentially Weighted Moving Average (EWMA) and Singular Spectral Analysis (SSA). We used nine wells to densify the velocity model of the Miranga Pole seismic cube and α = 0.025 for smoothing the slowness profile from the EWMA. We decomposed the slowness curve into three components with the SSA method, using only the first. The results, illustrated in temporal slices of the seismic cube, highlighted an improved resolution of the reflectors and the appearance of new reflections previously masked by the ineffective correction of the normal moveout. The third contribution is associated with noise attenuation. We developed a method that applies Singular Value Decomposition (SVD) in the time and frequency domain, using multidirectional spatial correlations in cross-spreads for ground roll attenuation. We used the seismic cube Polo-Miranga, due to its orthogonal geometry. Besides attenuating the ground roll, the method preserved, highlighted, and improved the lateral continuity of the seismic reflectors, increasing the signal-to-noise ratio of the data. The results are cross-spreads, inlines, and crosslines of the seismic cube.