Elastic wave least-squares reverse time migration based on optimized hybrid conjugate gradient method

Least-squares reverse time migration is a method of updating the reflection coefficient through fitting and multiple iterations based on reverse time migration, so as to improve the imaging quality to the greatest extent, so that the acquired imaging results have higher accuracy and better resolution. The traditional elastic wave least squares reverse time migration uses the conjugate gradient method or L-BFGS (Limited-memory BFGS) to update the reflection coefficient during the iteration process. The operation speed of the conjugate gradient method is slightly insufficient; and the L-BFGS method occupies a large amount of computing memory. This paper studies the least-squares reverse time migration based on the decoupled elastic wave equation, and updates the imaging value through a new hybrid conjugate gradient method based on memoryless variable BFGS (M-BFGS). The advantage of this method is that it can improve computational efficiency while ensuring imaging accuracy and occupying less memory. Compared with traditional methods, the advantage of this method is that it can improve computational efficiency while occupying less memory, and can also improve imaging accuracy with the same number of iterations.