A fast least-squares reverse time migration method using cycle-consistent generative adversarial network

With high imaging accuracy, signal-to-noise ratio, and good amplitude balance, least-squares reverse time migration (LSRTM) is an algorithm suitable for deep high-precision oil gas exploration. However, the computational costs limit its large-scale industrial application. The difference between traditional (RTM) LSRTM whether to eliminate effect of Hessian operator or not while solving matrix explicitly eliminating implicitly has a very requirement on computation storage capacity. We simulate inverse by training cycle-consistent generative adversarial network (cycleGAN) construct mapping relationship RTM results true reflectivity models. trained directly applied results, which improves quality significantly reducing calculation time. select three velocity models two respectively generate validation data sets, where involved in process. prediction sets show that with almost no additional effort. Finally, we apply only synthetics field data. predicted confirm effectiveness generalization proposed method.