It is of great significance to detect faults correctly in continental sandstone reservoirs in the east of China to understand the distribution of remaining structural reservoirs for more efficient development operation. However, the majority of the faults is characterized by small displacements and unclear components, which makes it hard to recognize them in seismic data via traditional methods. We consider fault detection as an end-to-end binary image-segmentation problem of labeling a 3D seismic image with ones as faults and zeros elsewhere. Thus, we developed a fully convolutional network (FCN) based method to fault segmentation and used the synthetic seismic data to generate an accurate and sufficient training data set. The architecture of FCN is a modified version of the VGGNet (A convolutional neural network was named by Visual Geometry Group). Transforming fully connected layers into convolution layers enables a classification net to create a heatmap. Adding the deconvolution layers produces an efficient network for end-to-end dense learning. Herein, we took advantage of the fact that a fault binary image is highly biased with mostly zeros but only very limited ones on the faults. A balanced crossentropy loss function was defined to adjust the imbalance for optimizing the parameters of our FCN model. Ultimately, the FCN model was applied on real field data to propose that our FCN model can outperform conventional methods in fault predictions from seismic images in a more accurate and efficient manner.

We have developed a migration scheme that can compensate absorption and dispersion with effective Q estimation and Fresnel zone identification based on deep learning. We use the U-Net neural network technology in deep learning to automatically identify Fresnel zones from compensated migrated dip-angle gathers and obtain the optimal aperture for migration, avoiding the tedious task of manually modifying the boundaries of Fresnel zones. Instead of the interval Q factor, we used an effective Q parameter to compensate absorption and dispersion. The effective Q is estimated using VSP well data and surface seismic velocity data. The proposed scheme can be incorporated into conventional seismic data processing workflow. A field data set was employed to validate the proposed scheme. Higher resolution imaging results with low noise levels are obtained.

The conventional time migration method does not consider the attenuation caused by the viscoelasticity of the underground media during the imaging process. Therefore, the final imaging amplitude and phase include inaccuracies caused by attenuation. In this study, we develop a migration scheme to compensate for absorption and dispersion using an effective quality factor (Q) estimation based on a back-propagation (BP) neural network. We use BP neural network technology to automatically estimate the effective Q value from stacked imaging data, thereby avoiding manual Q estimation using conventional methods. Our scheme can be incorporated into conventional seismic data-processing workflows. Furthermore, synthetic and field data sets are used to validate our scheme, which is used to acquire high-resolution images with low noise levels. In addition to developing a completely data-driven Q-value estimation strategy, this study demonstrates close integration of artificial intelligence, data mining, and conventional geophysics; our approach is appropriate for estimating the effective Q and has strong industrial application value and significance.

We have presented an inverse $Q$ filtering scheme that can compensate absorption and dispersion caused by intrinsic attenuation in subsurface media with a heterogeneous $Q$ model. We have adopted two methods to suppress high-frequency noise at the same time, one is to design a compensation operator with a fixed gain limit, and the other is to introduce an adaptive frequency-varying band calculation method. We use VSP data and seismic velocity data to estimate $Q$ model of the whole work area in a unique way. The proposed scheme can be incorporated into conventional seismic data processing workflow. Tests on synthetic and real data set demonstrate effectiveness of the proposed inverse $Q$ filtering.