A new correlation based least-squares reverse time migration

Least-squares reverse time migration (LSRTM) applies reverse time migration and demigration as migration and modeling operator to minimize the difference between the simulated and observed seismic data. We propose a new scheme for LSRTM and derive its corresponding conjugate gradient method to seek for the optimal image. By applying a correlation based objective function to LSRTM, we are able to emphasize not only on amplitude matching but also on phase information. The objective function incorporates the auto-correlation besides the cross-correlation. And it is the least-square of the two correlations. Numerical tests on synthetic models demonstrate that our new LSRTM scheme improves true amplitude imaging in the case of severely undersampled data and simultaneous source data.