Source-independent Least-squares Reverse Time Migration

Least-squares reverse time migration (LSRTM) has been proved to produce migration images with better quality than conventional migration. LSRTM is usually based on the assumption that the source wavelet is already known. To know the wavelet is difficult for the practical seismic explorations. We define the convolution-based objective function to perform the source-independent LSRTM. The examples of the conventional and source-independent methods of LSRTM with incorrect source wavelet prove that our method can eliminate the artifacts caused by the incorrect source wavelet. The new method not only recovers the reflectivity model more accurately, but also has faster convergence rate than the conventional method.