Seismic Impedance Inversion Based on Residual Attention Network

Deep learning (DL) has achieved promising results for impedance inversion via seismic data. Generally, these networks, composed of convolution layers and residual blocks, tend to deliver good results with deep architectures. Nevertheless, deep networks accompany a large number of parameters and longer training time. The volume of seismic data, especially 3-D scenarios, is very large. Therefore, it is particularly important to improve the accuracy while ensuring the model efficiency for practical implementation. With the flourishing new modules and techniques, DL has set the state of the art in many applications across a wide range of scientific and engineering disciplines. In this article, we present a residual attention network (ResANet), a convolutional neural network (CNN) incorporating with residual modules, and two attention mechanisms: channelwise attention and feature-map attention, for seismic impedance inversion. The proposed network can fuse multiscale channel information and recalibrate channelwise feature responses as well as receptive fields adaptively. At the same time, ResANet adopts grouped convolution, dilated convolution, and dropout techniques to improve the computation efficiency and stability. The Marmousi2 synthetic model and field data test results show that the proposed network outperforms several comparable neural networks in accuracy and generalization ability while ensuring efficiency for seismic data impedance inversion. For the field data test, transfer learning is also evoked to further improve the performance. ResANet tends to predict impedance with high resolution and strong lateral continuity compared with three closely related networks. The accuracy of ResANet is improved by 1–2 orders of magnitude on the six well logs provided in field dataset tests compared with commercial software (InverTrace Plus module in Jason) using the constrained sparse spike inversion (CSSI) method.