T033-0011
An Amplitude-preserving Adaptive Subtraction of Multiples for Deep Reflection Seismic Profile Based on Empirical Low-Rank Representation
An Amplitude-preserving Adaptive Subtraction of Multiples for Deep Reflection Seismic Profile Based on Empirical Low-Rank Representation
Friday, 11 December 2020
Poster
Abstract:
Deep reflection seismic profile technology is is known as on of the effective techniques to detect the lithosphere fine structure, which has the advantages of large detection depth, high resolution, accuracy and reliability. In recent years, the exploration of China's deep seismic reflection profile has gradually expanded to offshore waters. Compared with land data, marin data contains different types of multiples (surface-related multiples, internal multiples, etc.). These multiples are a kind of coherent noise, which interfere with the deep events and increase the difficulty of subsequent seismic interpretation. Furthermore, the deep reflection seismic profile has a longer observation system, a longer sampling time, and a weaker deep events. Therefore, how to maintain deep evemts while suppressing multiples is the key technology in the offshore waters.
The SRME (surface-related multiple elimination) method widely used in the industry, which achieves multiple attenuation through multiple prediction and adaptive subtraction. However, the conventional adaptive subtraction method based on the principle of minimum energy is not applicable at the intersection of primary and multiple, and the phenomenon of primary damage and multiple residuals will reduce the accuracy of deep reflection seismic profile. Based on previous research, we propose an amplitude-preserving multiple subtraction method based on empirical low-rank representation. Firstly, we use empirical mode decomposition method to improve the conventional low-rank representation method. By adaptively decomposing the seismic signal into a low-rank subset with high signal-to-noise ratio, simple inclination, and smooth phase of events, we optimize the parameters selection of the local window during the low-rank representation, which means reducing the complexity of the dip component of the seismic signal. Then we adopt the conventional adaptive subtraction method in different low-rank subsets to avoid the event intersection in the adaptive subtraction, and reconstruct each subset to preserve the energy of deep events during multiple attenuation. The proposed method provides technical support for deep reflection seismic profile.
The SRME (surface-related multiple elimination) method widely used in the industry, which achieves multiple attenuation through multiple prediction and adaptive subtraction. However, the conventional adaptive subtraction method based on the principle of minimum energy is not applicable at the intersection of primary and multiple, and the phenomenon of primary damage and multiple residuals will reduce the accuracy of deep reflection seismic profile. Based on previous research, we propose an amplitude-preserving multiple subtraction method based on empirical low-rank representation. Firstly, we use empirical mode decomposition method to improve the conventional low-rank representation method. By adaptively decomposing the seismic signal into a low-rank subset with high signal-to-noise ratio, simple inclination, and smooth phase of events, we optimize the parameters selection of the local window during the low-rank representation, which means reducing the complexity of the dip component of the seismic signal. Then we adopt the conventional adaptive subtraction method in different low-rank subsets to avoid the event intersection in the adaptive subtraction, and reconstruct each subset to preserve the energy of deep events during multiple attenuation. The proposed method provides technical support for deep reflection seismic profile.