Multisource least-squares extended reverse time migration with preconditioning guided gradient method

Abstract

Least-squares reverse-time migration (LSRTM) provides image of reflectivity with high resolution and compensated amplitude, but the computational cost is extremely high. One way to improve efficiency is to encode all of shot gathers into one or several super-shot gathers with designed encoding functions so as to solve a smaller number of wave-equations at each iteration. Another way is to use different kinds of preconditioners to accelerate the convergent rate of iterations. In this abstract, we combine these two methods together and extend them into prestack imaging, which we call extended reverse-time migration (ERTM). The sparsity of reflectivity is used as prior information and preconditioning guided gradient (PGG) method is developed to suppress crosstalk introduced by phase encoding and improve resolution of the extended image in the subsurface offset domain. Numerical tests on SEG/EAGE salt model show that our proposed method can provide a more reliable extended image with almost the same cost of ERTM, which makes the method valuable in migration velocity analysis and AVO/AVA analysis.