3-D Seismic Multi-horizon Extraction Based on a Domain Adaptive Deep Neural Network

Xin He,Yifeng Fei,Cheng Zhou,Feng Qian,Yaojun Wang,Gang Yu,Guangmin Hu
DOI: https://doi.org/10.1109/tgrs.2024.3472120
IF: 8.2
2024-01-01
IEEE Transactions on Geoscience and Remote Sensing
Abstract:3-D seismic multi-horizon extraction is crucial for 3D sequence stratigraphy analysis and reservoir modeling. Deep neural networks (DNNs) often cause dislocated horizons in regions with complex geological structures, such as faults and unconformities. This issue arises from two main factors. First, obtaining horizon labels from real seismic data is subjective and expensive, resulting in existing DNNs lacking field seismic data labels, which limits their ability to extract multiple horizons across complex geological structures. Second, directly extracting multiple horizons using only seismic data reduces precision in discontinuous areas with faults and unconformities. To address these issues, this paper proposes a domain adaptation layer based on multi-kernel maximum mean discrepancy (MK-MMD) and designs a domain adaptive deep neural network (DA-DNN) for seismic multi-horizon extraction. We map synthetic and field seismic data to the reproducing kernel Hilbert space (RKHS) and use MK-MMD to minimize feature differences between them. Unlike the traditional multi-scale Gaussian kernel function used in MK-MMD, this paper constructs a hybrid kernel function that integrates multi-scale Gaussian and multi-scale Laplacian kernels. The multi-scale Gaussian kernel evaluates local-to-global feature differences in continuous areas, whereas the multi-scale Laplacian kernel captures rapid feature variations in complex geological structures. Finally, a few seismic horizons and fault attributes guide the training process of DA-DNN, further improving the prediction accuracy of multi-horizon extraction. Synthetic and field seismic examples show our model can extract seismic multiple horizons more accurately in field seismic data and performs better in discontinuous areas.
What problem does this paper attempt to address?