A 3D inversion based Least-Squares Reverse Time Migration (LSRTM) technique was developed. The algorithm uses the RTM as the forward modeling and inversion engine to minimize the amplitude differences between the observed data and the synthetic modeled data. In turn, the final LSRTM will deliver the reflectivity model that will generate the true corresponding amplitude; the migration artifacts are suppressed as well since they are not contained in the observed field data. Compared with the initial RTM image, the LSRTM images from the synthetic data and field data examples show the improved amplitude response and higher resolution gained by suppressing migration artifacts and sharpening the subsurface reflectors' reflectivity.

High-resolution seismic imaging is crucial for deriving an accurate geological section and ensuring successful petroleum exploration. However, traditional high-resolution imaging methods such as least-squares reverse time migration (LSRTM) and migration deconvolution usually involve high computational costs for approximating the inverse Hessian matrix with limited improvements in resolution. To obtain high-resolution migration images and reduce computing costs, we propose a deep learning-based point spread function (PSF) deconvolution method. We decompose the large Hessian matrix into small dispersed PSFs and design a convolutional neural network (CNN) to automatically predict the deconvolution operator of every single PSF. The deconvolution operator eliminates the PSF effect and balances the insufficient illumination of the acquisition geometry. To obtain an efficient CNN model for predicting deconvolution operators, we trained 2,500 pairs of PSFs and their corresponding deconvolution operators collected from part of a model using the regular PSF deconvolution method. After training the network, we could predict all deconvolution operators in a few seconds for the following processing of migration images. The results on synthetic and field-data applications indicate that our method could provide a deblurred migration image comparable to that of the LSRTM approaches with significantly reduced computational and memory costs.

No abstract

No abstract

No abstract