AVO Inversion for Low-Frequency Component of the Model Parameters Based on Dual-Channel Convolutional Network

Amplitude variation with offset (AVO) inversion is an important method for estimating elastic parameters in geosciences. The inversion results are highly affected by the initial low-frequency model. Deep learning, as a data mining algorithm, has the potential to capture more reliable low-frequency components from seismic data and well logs. In order to fully utilize limited seismic data and recover more reliable low-frequency components of elastic parameters, a data-driven complex frequency-domain AVO inversion is proposed by combining a convolutional neural network (CNN) with a complex frequency forward solver. The proposed method makes full use of the advantages of deep learning algorithms and the low-frequency components of the attenuation wave field in the complex frequency domain. To better fit the characteristics of complex frequency-domain seismic data, we designed a dual-channel U-shaped CNN (DC-UCNN) as an inversion network to extract the low-frequency components contained in the real and imaginary parts of the attenuated wave field. Furthermore, a data-driven complex frequency forward modeling operator is used to constrain the training of the inversion network to improve its reliability. The experiment on the Marmousi2 model shows that the proposed DC-UCNN-based complex frequency-domain inversion has better accuracy than the traditional complex frequency-domain AVO inversion and UCNN (U-shaped CNN)-based time-domain inversion. Finally, the application of field data has proven the feasibility of the proposed method, which can recover richer and more reliable low-frequency components far from the well location.