|
| | Abstract | | Interpreting seismic data at subduction zones need experiences to determine the complex geological structures and induce the down-slope transport of large masses (i.e., slumps) which may cause many earthquakes. Recently, Artificial Intelligence (AI) has provided easy and high effectiveness solutions for 2D and 3D data analysis such as colored images and face recognition. Therefore, this paper proposes the use of AI (convolutional neural networks CNNs) to automatically detected multiples geological features such as slumps and thrust faults. In this paper, we apply our CNNs on 3D seismic data acquired in the Nankai subduction zone. The training was selected from real 3D seismic data to achieve 98.8% classification accuracy for slump units. Our result was compared with previously published results and it showed high similarity. We also interpret thrust faults using CNNs and compare our approach with other methods. CNNs shows superiority on other methods and a high spatial resolution. |
|
|
|
| |
|